ANNUAL PROGRESS IN CHILD PSYCHIATRY AND CHILD DEVELOPMENT 2000–2001
ANNUAL PROGRESS IN CHILD PSYCHIATRY AND CHILD DEVELOPMENT 2000– 2001 Edited by
MARGARET E.HERTZIG, M.D. Professor of Psychiatry Cornell University Medical College and ELLEN A.FARBER, Ph.D. Clinical Assistant Professor of Psychology Cornell University Medical College
Brunner-Routledge New York and Hove
Published in 2003 by Brunner-Routledge 29 West 35th Street New York, NY 10001 www.brunner-routledge.com Published in Great Britain by Brunner-Routledge 27 Church Road Hove, East Sussex BN3 2FA www.brunner-routledge.co.uk Copyright © 2003 by Taylor & Francis Books, Inc. Brunner-Routledge is an imprint of the Taylor & Francis Group. This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk”. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. 10 9 8 7 6 5 4 3 2 1 Library of Congress Cataloging-in-Publication Data is available from the publisher. ISBN 0-203-44952-5 Master e-book ISBN
ISBN 0-203-45789-7 (Adobe e-Reader Format) ISBN 0-415-93548-2 (Print Edition)
CONTENTS Introduction I. DEVELOPMENTAL ISSUES 1. Child Development and the PITS: Simple Questions, Complex Answers, and Developmental Theory Frances Degen Horowitz 2. Developing Mechanisms of Self-Regulation Michael I.Posner and Mary K.Rothbart 3. Implications of Attachment Theory for Developmental Psychopathology L.Alan Sroufe, Elizabeth A.Carlson, Alissa K.Levy, and Byron Egeland 4. Attachment Security in Infancy and Early Adulthood: A Twenty-Year Longitudinal Study Everett Waters, Susan Merrick, Dominique Treboux, Judith Crowell, and Leah Albersheim 5. Behavioral and Physiological Responsivity, Sleep, and Patterns of Daily Cortisol Production in Infants with and without Colic Barbara Prudhomme White, Megan R.Gunnar, Mary C.Larson, Bonny Donzella, and Ronald G.Barr 6. Imaginary Companions of Preschool Children Tracy R.Gleason, Anne M.Sebanc, and Willard W.Hartup II. PARENTING 7. Contemporary Research on Parenting: The Case for Nature and Nurture W.Andrew Collins, Eleanor E.Maccoby, Laurence Steinberg, E.Mavis Hetherington, and Marc H.Bornstein 8. Social versus Biological Parenting: Family Functioning and the Socioemotional Development of Children Conceived by Egg or Sperm Donation Susan Golombok, Clare Murray, Peter Brinsden, and Hossam Abdalla 9. Parenting among Mothers with a Serious Mental Illness Daphna Oyserman, Carol T.Mowbray, Paula Allen Meares, and Kirsten B.Firminger III. ATTENTION DEFICIT-HYPERACTIVITY DISORDERS 10. Diagnostic Efficiency of Neuropsychological Test Scores for Discriminating Boys with and without Attention Deficit-Hyperactivity Disorder
viii
3 19 39
57
66 91
113
141
160
11.
12.
13.
Alysa E.Doyle, Joseph Biederman, Larry J.Seidman, Wendy Weber, and Stephen V.Faraone Efficacy of Methylphenidate among Preschool Children with Developmental Disabilities and ADHD Benjamin L.Handen, Heidi M.Feldman, Andrea Lurier, and Patty Jo Huszar Murray ADHD in Girls: Clinical Comparability of a Research Sample Wendy S.Sharp, James M.Walter, Wendy L.Marsh, Gail F.Ritchie, Susan D.Hamburger, and F.Xavier Castellanos Stimulant Treatment for Children: A Community Perspective Adrian Angold, Alaattin Erkanli, Helen L.Egger, and E.Jane Costello
IV. OTHER CLINICAL ISSUES 14. The Altering of Reported Experiences Daniel Offer, Marjorie Kaiz, Kenneth I.Howard, and Emily S.Bennett 15. Developmental Coordination Disorder in Swedish 7-year-old Children Björn Kadesjö and Christopher Gillberg 16. Thirty-Three Cases of Body Dysmorphic Disorder in Children and Adolescents Ralph S.Albertini and Katharine A.Phillips 17. Case Series: Catatonic Syndrome in Young People David Cohen, Martine Flament, Pierre-Francois Dubos, and Michel Basquin 18. Toward a Developmental Operational Definition of Autism Jane E.Gillham, Alice S.Carter, Fred R.Volkmar, and Sara S.Sparrow 19. Adolescent Onset of the Gender Difference in Lifetime Rates of Major Depression: A Theoretical Model Jill M.Cyranowski, Ellen Frank, Elizabeth Young, and M.Katherine Shear 20. Social/Emotional Intelligence and Midlife Resilience in Schoolboys with Low Tested Intelligence George E.Vaillant and J.Timothy Davis V. TREATMENT ISSUES 21. Effectiveness of Nonresidential Specialty Mental Health Services for Children and Adolescents in the “Real World” Adrian Angold, E.Jane Costello, Barbara J.Burns, Alaattin Erkanli, and Elizabeth M.Z.Farmer 22. Early Intervention Programs for Children with Autism: Conceptual Frameworks for Implementation Heather Whiteford Erba 23. Treatment for Sexually Abused Children and Adolescents Karen J.Saywitz, Anthony P.Mannarino, Lucy Berliner, and Judith
197
222
236 251
272 288
305
318 331
349
364
382
395
24.
25.
A.Cohen Neuroleptic Malignant Syndrome in Children and Adolescents Raul R.Silva, Dinohra M.Munoz, Murray Alpert, Ilisse R.Perlmutter, and Jose Diaz AHA Scientific Statement: Cardiovascular Monitoring of Children and Adolescents Receiving Psychotropic Drugs Howard Gutgesell, Dianne Atkins, Robyn Barst, Marcia Buck, Wayne Franklin, Richard Humes, Richard Ringel, Robert Shaddy, and Kathryn A.Taubert
VI. SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION 26. Twenty Years’ Research on Peer Victimization and Psychosocial Maladjustment: A Meta-Analytic Review of Cross-Sectional Studies David S.J.Hawker and Michael J.Boulton 27. Charting the Relationship Trajectories of Aggressive, Withdrawn, and Aggressive/Withdrawn Children During Early Grade School Gary W.Ladd and Kim B.Burgess 28. Violent Behavior in Children and Youth: Preventive Intervention from a Psychiatric Perspective Group for the Advancement of Psychiatry, Committee on Preventive Psychiatry 29. Agents of Change: Pathways through which Mentoring Relationships Influence Adolescents’ Academic Adjustment Jean E.Rhodes, Jean B.Grossman, and Nancy L.Resch 30. Initial Impact of the Fast Track Prevention Trail for Conduct Problems: II. Classroom Effects Conduct Problems Prevention Research Group Permissions
415
435
450
461
488
521
534
553 574
INTRODUCTION This millennial edition of the Annual Progress in Child Psychiatry and Child Development includes papers published in 1999 and 2000. The collection provides an overview of the wide-ranging interests of both researchers and clinicians at the turn of the twentieth century. Thirty papers, organized into six sections: Developmental Issues; Parenting; Attention Deficit Hyperactivity Disorder; Other Clinical Issues; Treatment Issues; and Violence and Victimization include both research reports and reviews of the literature.
DEVELOPMENTAL ISSUES The six papers in this section cover a wide range of topics. The paper by Frances D.Horowitz, written as the presidential address to the Society for Research in Child Development, was a warning to researchers about how to present child development knowledge to the public. She used the acronym “PITS” (person in the street) to refer to the public. Horowitz was concerned that developmental scientists might oversimplify child development in an attempt to respond to public interest. For example, if one were asked whether child care had a negative or positive effect on development one should not give a yes/no response. She presents a model that highlights the range of experiences (environmental and relationship factors) that interact with constitution to result in developmental outcomes. The second paper by Posner and Rothbart is an intriguing forward-looking discussion of the topic of self-regulation. The authors refer to studies of the self-regulation of attention and cognitive processes. The paper highlights recent work in neuroimaging that allows us to study brain plasticity and examine individual differences in self-regulation. The authors believe that individual differences in effortful control have implications for normal and pathological development. The next two papers are on attachment. The paper by Sroufe and colleagues reviews “implications of attachment theory for developmental psychopathology.” Attachment categories typically have been proposed to be variations on normal rather than pathological development. The authors elaborate on the transactional model for understanding the various pathways underlying pathology and the role of attachment relations within that model. Waters and his colleagues present new data on the stability of attachment classifications. They assessed 50 21-year-old individuals who had been seen in the Strange Situation when they were 12 months of age. The young adults completed the Adult Attachment Interview. There was a significant correspondence of categories between age one and age twenty-one. This is a nice, clean presentation of stability in a middle-class sample. The next paper by White and colleagues is an interesting study of an important topic,
the infant with colic. The authors used parent diaries and physiological measures. The data suggested that colic might be associated with a delay in the establishment of the circadian rhythm. Colic, although only lasting for a few months, can present significant burdens on family relationships. An understanding of the physiological basis of colic will one day lead to meaningful interventions. The last paper in this section is a small data-gathering study of an intriguing topic. Imaginary friends have long been noted in child development literature. However, there have been little data to suggest which children develop imaginary friends and whether the children who have imaginary friends have socialization difficulties. Gleason and colleagues distinguish children with invisible friends, personified objects, and with no imaginary companions.
PARENTING Part II presents three papers on parenting. The first paper, “Contemporary research on parenting: The case for nature and nurture,” is a collaborative effort by five psychologists, all well-established researchers on parenting. This is a thoughtful presentation of the different contexts for studying parental influence on child development. These include behavior genetics models that tease apart the percentage of nature and nurture on specific characteristics, gene-by-environment models, and parental influence models. They also present a discussion of environmental influences such as peers and neighborhood that affect parental influence. The second paper by Golombok and Murray and colleagues, discusses issues in parenting created by new reproductive technologies. They investigated parental and child functioning in four groups of families in which the parentchild genetic link varied. The groups included adoptive families, families created by in vitro fertilization, egg donation families, and donor sperm families. It is only since the 1980s that woman can have donor eggs implanted and conceive a child not genetically linked to them. Results are counterintuitive and worthy of continuing investigation. The paper raises interesting issues, including telling children about genetic parentage. The third paper is, “Parenting among mothers with a serious mental illness.” Improvements in medication, deinstitutionalization, and communitybased support programs have resulted in more women with serious mental illness being able to have and raise children. Oyserman and her colleagues review an important literature. They describe what is known about mothers with serious mental illness, including schizophrenia and affective disorders.
ATTENTION DEFICIT-HYPERACTIVITY DISORDER The diagnosis and treatment of attention deficit-hyperactivity disorder (ADHD) is the subject of increasing interest and concern among those devoted to enhancing the welfare of children, including parents, pediatricians, child psychiatrists and psychologists, as well as educators. Debate, often acrimonious, has tended to focus on the validity of diagnosis
and the appropriateness of various treatment options, most particularly pharmacological interventions. The four papers in this section, which provide a sampling of recent research, expand the empirical base for clinical decision making. As Doyle, Biederman, Seidman, Webber, and Faraone indicate in the first paper in this section, clinicians have long speculated that neuropsychological tests could aid in the diagnostic classification of children with ADHD. However, the results of this important study clearly demonstrate the limitations that attach to efforts to use neuropsychological test results to “validate” the diagnosis of ADHD. Careful history taking, clinical interviewing, and reports of informant observations of behavior, provide the basis for clinical diagnosis. Neuropsychological tests of attention and executive functioning under-identify children who exhibit abnormal patterns of inattention, impulsivity, and hyperactivity. Nevertheless, by clarifying patterns of strengths and weaknesses, neuropsychological testing may usefully supplement educational and treatment planning for individual children. The study reported by Handen, Feldman, Lurier, and Murray is a small N study, but nevertheless, it provides much needed information regarding the effects and side effects of methylphenidate (MPH) in preschool children with ADHD and developmental disabilities. The issue of medicating preschool children with developmental disabilities and/or mental retardation is controversial because it is often difficult to determine whether activity level and impulsive style are in excess of what would be expected for mental age. Nevertheless as the authors emphasize, the ADHD-like behaviors exhibited by at least some developmentally disordered preschoolers may result in suspension from intervention programs as well as severe family disruption. The results suggest that preschool children with developmental disabilities respond to MPH at rates similar to those of typically developing children, but might well be at greater risk of developing significant adverse side effects. The authors provide useful guidelines for noting MPH effects and side effects during standard office visits via observations of play and motherchild interactions. In the third paper in this section, Sharp, Walter, Marsh, Ritchie, Hamburger, and Castellanos describe the clinical characteristics and response to stimulant medication of girls with ADHD. As these authors point out, the investigation of ADHD in girls poses complex questions of referral bias and selection criteria, and almost all research has focused exclusively on boys. The clinical and medication response of this sample of girls proved to be strikingly similar to a previously studied research sample of boys with ADHD. The final paper in this section provides a community perspective on the use of stimulant medication for the treatment of ADHD. The study by Angold, Erkanli, Egger, and Costello examines the use of prescribed stimulants in relation to research diagnoses of ADHD in a community sample of children living in an 11-county region in the Great Smoky Mountains. Data on a representative sample of 4,500 children and adolescents aged 9, 11, and 13 years were collected in four waves between 1992 and 1996. The results of this study provide evidence for both the over- and under-treatment of symptoms of ADHD. Although the authors conclude that stimulant treatment was being used in ways substantially inconsistent with current diagnostic guidelines in this geographic area, it should be noted that the extent of inappropriate treatment is difficult to determine
because the diagnoses were based on parental report alone. Nevertheless, the results underscore how important it is for every clinician to establish clear indications for stimulant treatment on a case-by-case basis, to monitor the effectiveness of treatment in reducing target symptoms on an ongoing basis and discontinue treatment in a timely manner if clearly defined treatment goals are not met.
OTHER CLINICAL ISSUES The seven papers in this section address a range of topics of interest to practicing clinicians. In the first report, Offer, Kaiz, Howard, and Bennett direct attention to how reported experiences may be altered. The authors utilized data from a longitudinal study of development from adolescence to middle adulthood of a group of essentially normal males to investigate the stability of memory concerning perceptions of events and relationships that had occurred during adolescence. The findings indicate that the degree of concordance between responses made during adolescence and adulthood was no better than would be expected by chance. Whether these findings of this study of normal mentally healthy individuals can be extended to reports of past experiences of those who are medically or psychiatrically ill remains to be determined. Nevertheless, these results direct attention to the importance of the use of collateral sources to enhance the validity of relevant historical information when assessing adults and older adolescents. That some children lack the motor skills required for everyday activities is well known, but the prevalence, comorbidity, and outcome of developmental coordination disorder (DCD) is less clear. Kadesjo and Gillberg remedy this gap in their report of a population study of Swedish 7-year-old children, who were followed up at yearly intervals for an additional three years. DCD is a common problem, frequently comorbid with ADHD. Clinicians are urged to fully assess motor clumsiness, particularly in children with ADHD, to provide a basis for the development of treatment plans to minimize the functional impact of DCD. Although body dysmorphic disorder (BDD), a preoccupation with a nonexistent or slight defect in appearance, usually begins during adolescence, the disorder has been little studied in this age group. In the third paper in this section Albertini and Phillips report on the demographic characteristics, phenomenology, associated psychopathology, and treatment history and response of 33 children and adolescents below the age of 17, meeting DSM-IV criteria for the diagnosis of BDD. The results of this carefully executed descriptive study clearly indicate that social impairment is nearly universal among children and adolescents with BDD. Although surgical, dermatological, and dental procedures appear ineffective, serotonin reuptake inhibitors (SRIs) often appear to be helpful, even if the symptoms are delusional. The wealth of clinical information in this paper will aid clinicians to recognize and more effectively treat this distressing, often secret and underrecognized, but very debilitating condition. In the next paper in this section, Cohen, Flament, Dubos, and Basquin direct attention to catatonia, an infrequent but severe condition in young people. The authors’ augment their review of all reports of catatonic adolescents found in the literature between 1977 and 1997 (boys=25, girls=22) with an additional 9 consecutive cases seen during the past
6 years. The clinical features of catatonia as well as etiologies, complications, and treatment are similar to those reported in the adult literature. However, unlike adult studies, in which catatonic women represent 75% of the cases, of the total of 51 children and adolescents summarized in this report 60% were male. Although schizophrenia appears to be the most frequent associated diagnosis in adolescents, comorbid organic conditions, mood disorders, and developmental disorders have also been described. This paper provides clinicians with crucial information necessary for the diagnosis and treatment of this rare but sometimes fatal condition as it presents in children and adolescents. In their paper, “Toward a developmental operational definition of autism,” Gillham, Carter, Volkmar, and Sparrow point out that traditional diagnostic schemes typically list symptoms, but provide little guidance on how to incorporate information about developmental level in making a diagnosis. This study explores the ability of the Vineland Adaptive Behavior Scales to identify children with autism. The findings suggest that evaluating children’s socialization in the context of their developmental level may allow for a better appreciation of the ways in which symptoms change with age, and may reduce confusion about diagnosis among clinicians and parents. Depression is the next clinical issue addressed in this section. In it Cyranowski, Frank, Young, and Shear offer a theoretical model for understand-ing the adolescent onset of the gender difference in lifetime rates of major depression. In it the authors bring a vast array of information to bear as they outline the affiliative proclivities and potential depressogenic vulnerabilities of adolescent girls. As the authors indicate, the proposed theoretical framework has considerable heuristic value because it connects multiple fields of inquiry pointing the way to further programmatic research. The final paper in this section examines social/emotional intelligence and midlife resilience in schoolboys with low tested intelligence. As Vaillant and Davis point out, the natural history of low intelligence has been largely ignored, and the findings of this study will shatter many stereotypes. When the adult adjustment of 73 inner-city boys with a mean IQ of 80, followed prospectively from age 14 to 65 were compared with a socioeconomically matched sample of 38 boys with a mean IQ of 115, half of the low-IQ subjects were found to enjoy incomes as high and had children as well-educated as did highIQ men. Moreover the resilient low-IQ men were more likely to be generative, use mature defenses, and enjoy warm object relations than the high-IQ group as a whole. The authors provide several case vignettes to illustrate the proposition that “during the school years, low tested IQ is a terrible curse, but IQ is not destiny.” Resilience with respect to low tested intelligence appears to be mediated by those social skills often termed “emotional intelligence,” which includes the ability to know and manage feelings.
TREATMENT ISSUES The five papers in this section examine the current status of a range of treatments for psychiatrically and developmentally disordered children and adolescents. In the first paper, Angold, Costello, Burns, Erkanli, and Farmer provide an epidemiological perspective on the effectiveness of psychiatric treatment in the “real world.” As these
authors note, efficacy studies usually take place in resource-rich, university-based settings and involve highly selected, often nonreferred subjects who are willing to be randomized and who have relatively homogeneous histories. Whether such treatments are effective in the real world of the mental health practitioner’s office is clearly an important question—and the results of this study provide some answers. The use of a large, longitudinal, nonexperimental, community-based study to explore treatment effects naturalistically trades experimental rigor for ecological validity. Within these constraints, however, the results do suggest that the mental health care system for children can reduce symptoms in the real world, provided that the length of treatment is of sufficient length to be efficacious. Because real improvement does not become apparent until an individual has received more than 8 sessions, it is incumbent upon both individual practitioners and service planners to be cognizant of this important finding. Although there is a growing consensus that the functional potential of children with autism can be increased following exposure to intensive early intervention programs, evidence regarding the differential effectiveness of different programmatic contents is less clear. Nevertheless, families, physicians, and other service providers require information in order to make informed decisions regarding various intervention strategies. In the second paper in this section, Erba describes four diverse early intervention programs—discrete trial training, LEAP, floor time, and TEACCH, for children with autism. The theoretical underpinnings, intervention procedures, and connections between theory and practice are illustrated and available research outcomes are discussed. The paper is a useful resource for the practitioner, providing guidance to the family of a recently diagnosed young autistic child. Saywitz, Mannarino, Berliner, and Cohen provide an encyclopedic and very thoughtful review of research demonstrating the variable effects of childhood sexual abuse, the need for intervention, and the effectiveness of available treatment models. The authors emphasize that there is no evidence of a single cohesive syndrome resulting from child sexual abuse, although more than 50% of sexually abused children meet partial or full criteria for post-traumatic stress disorder. This review explicates well-controlled outcome studies that provide empirical evidence for extending and modifying treatment models from mainstream clinical practice to sexually abused children. Sensationalistic fringe treatments that treat sexually abused children as a special class of patients are not considered. A continuum of interventions, ranging from psychoeducation and screening, to short-term, abuse-focused, cognitive-behavioral therapy with family involvement, to more comprehensive long-term plans for multiproblem cases are considered. This review provides an invaluable guide, both to those who specialize in the area of the assessment and treatment of children who have been sexually abused as well as the practitioner seeking guidance regarding a rarely encountered individual case. The final two papers in this section summarize currently available information regarding serious and potentially fatal adverse effects of psychotropic medication. Neuroleptic malignant syndrome (NMS), a cluster of symptoms including hypertonicity, autonomic instability, fever, and cognitive disturbance, occurs in 0.5% to 1.4% of all individuals exposed to neuroleptics. Silva, Munoz, Alpert, Perlmutter, and Diaz summarize what is known about this condition as it occurs in children and adolescents. Their review of 77 reported cases of NMS provides a guide to early detection and
treatment interventions that can reduce the mortality and morbidity of this serious iatrogenic condition. The official scientific statement of the American Heart Association summarizes the cardiovascular effects of commonly used psychotropic medications in children and adolescents. The review—which details drug effects on the electrocardiogram, summarizes potentially dangerous drug interactions, and provides recommendations for cardiovascular monitoring—was prompted by reports of sudden deaths in children and adolescents receiving psychotropic medication. Its several tables and clearly stated recommendations make muchneeded information readily available to the prescribing practitioner.
SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION This section contains five papers. In the past few years, there have been several highly publicized cases of school violence. These incidences of children killing other children as well as teachers were exceptionally dramatic because they occurred in “unexpected” places, such as suburban and rural schools rather than urban, inner-city schools. These incidents have led more child professionals to try to understand which children are at risk for violence and what interventions might be successful at curbing violence. The first paper in this section is a review of cross-sectional studies of peer victimization on children’s adjustment. The meta-analysis by Hawker and Boulton indicated that children who are bullied by others tend to be high on psychosocial distress, particularly depression and loneliness. The second paper presents two years of longitudinal data highlighting the early behavioral characteristics that place children at risk for maladjustment and relationship problems. Ladd and Burgess present data from the Pathways Project, a multisite study. Teacher ratings of children’s aggressive and antisocial behaviors in the fall of kindergarten were used to place children into one of four behavioral risk groups. The trajectories of these groups were studied through the second grade in the current paper. The withdrawn group in kindergarten did not have relationship difficulties in the early years. Aggressive children had relationship difficulties with peers and teachers. The aggressive/ withdrawn children had the most difficulty. They were most often friendless, victimized, and dissatisfied. This study highlights the stability of behavioral risk and maladaptive relationships in the early grade school years. The third paper is a position paper by the Group for the Advancement of Psychiatry. Based on a literature review, the committee presented rates of violence and identified biological, psychological, and sociocultural factors that increase the risk of perpetrating violence. Because attempts to treat violence on an individual basis are costly and of limited success, the group advocates a tripartite model of prevention. This includes introducing universal measures to affect populations, selective measures designed for atrisk groups, and individual measures. In sum, the authors contend that violence is not really a sudden event. Its associated psychopathology develops slowly and has numerous psychiatric markers, thus putting child mental health practitioners in a position of early identification and intervention.
Numerous interventions to reduce violence and victimization have been developed and are being tested to determine their effectiveness. These include anti-bullying and conflict resolution curricular. The last two papers describe interventions. One is an individual intervention using mentors and the other is a “universal” classroom-based intervention. Rhodes, Grossman, and Resch studied 959 youth who applied to the Big Brother programs. Previous studies have found reduced rates of juvenile delinquency and substance abuse. Approximately half of the sample was placed with mentors and the remainder was used as controls. The children completed baseline and follow-up interviews 18 months later. This study highlights that a consistent mentoring relationship can serve as a corrective emotional experience for youth who have had unsatisfying relationships with their own parents. A positive mentoring experience was associated with improved parental relationships that in turn led to improved self-worth and achievement. The last paper is an interesting example of using entire schools to introduce interventions aimed at reducing violence. The Conduct Problems Prevention Research Group funded by NIMH and others is doing multi-site longitudinal studies. Someone interested in the topic should see other papers by members of this group. This paper describes the method of combining a universal intervention, the Fast Track PATHS (Promoting Alternative Thinking Strategies) classroom curriculum with a selective intervention. The selective intervention involved identifying the 10% of children with the highest incidence of early conduct problems and offering weekly parenting support classes, small-group social skills interventions, academic tutoring, and home visits. This intensive intervention was deemed necessary to reduce risk factors and promote protective factors in children on the path for antisocial behavior. It was believed that providing services to the classrooms would allow children to generalize their newly learned skills in a supportive setting. Promoting the development of social competence in all children should improve classroom atmosphere and social relations for all. The classroom was the unit of analysis in this paper and there is an interesting discussion of teachers’ interest and ability to implement interventions. This is an early paper from an ongoing study that will be of interest to many in education and school-age intervention. In summary, the papers included in this edition of the Annual Progress in Child Development and Child Psychiatry cover a multiplicity of topics. Some serve to deepen our understanding of the diagnosis and treatment of long recognized and well-described clinical phenomena, others direct attention toward issues of broader social concern, whereas still others focus normal developmental processes from both a neural and behavioral perspective—and as such, reflect the state of research interest and clinical practice at the turn of the century.
Part I DEVELOPMENTAL ISSUES
PART I: DEVELOPMENTAL ISSUES
1 Child Development and the PITS: Simple Questions, Complex Answers, and Developmental Theory Frances Degen Horowitz
The enormous popular interest in the field of child development makes it incumbent upon developmental scientists to convey with care the complexity of development lest oversimplified popular accounts gain credibility. Recent attempted models of development do include the range of variables and complexities that need to be accommodated in accounting for development. A model is presented here that incorporates many of the elements of recent models but elaborates on the role of experience in relation to the constitutional, cultural, economic, and social factors that contribute to advantages and disadvantages in children’s development. The importance of accommodating data from prior theoretical perspectives and the importance of the contributions from neuroimaging studies are discussed as they are critical for successful theory building in the field of child development.
INTRODUCTION For those who have not yet heard or figured it out, “Child Development and the PITS” translates into Child Development and the Person in the Street—the person in the street who asks simple questions and wants simple answers, who is puzzled by complex responses, and is terribly impatient with the nuances and qualifications that characterize contemporary theories of development. Some of you might have thought PITS was a reference to William James’ “TCPITS,” the common people in the street, but I actually modeled it on the title of a 1940s book on symbolic and mathematical logic by Lillian Lieber, MITS, WITS, and Logic (Lieber, 1960). MITS is the Man-In-The-Street and WITS is the Woman-In-The-Street. That slim volume, in its several editions, was and is a clever and sometimes humorous attempt to convey the essential aspects of symbolic logic to the person in the street. 1999 Presidential address to the Society for Research in Child Development.
Annual progress in child psychiatry and child development 2000–2001
4
Now if symbolic and mathematical logic for the man and woman in the street was a novelty 50 years ago, not so for child development. From the beginning of the modern serious focus on the study of children, well before the founding of the Society for Research in Child Development in 1933, surely dating back at least to the early days of the child study movement in the 1880s, popularized information about children and their development was aimed at people in the streets—at mothers and fathers and those responsible for the health and welfare of children (Cairns, 1983; Sears, 1975; Senn, 1975). And certainly, throughout the twentieth century, there has been no dearth of welland ill-informed books advising parents on the care of infants and children; wonderful and sometimes scary admixtures of well-grounded evidence and passionate advocacy. And it continues, increasing geometrically. Hit “parenting” at Amazon.com and one can browse the 75 bestsellers under the general title of parenting and families, or the 75 bestsellers on discipline, or on emotions and feelings, or on morals and responsibility. In the 12 pages that you can print out listing the 75 bestsellers on parenting and families you will note a number of volumes written by members of our Society along with the old standards—Spock’s Baby and Child Care (Spock, 1998)—as well as recent books of advice on raising the spirited child, the strong-willed child, the emotionally intelligent child, the nonconforming child, and the happy child. For the web sophisticate there is the National Parent Information Network (www.npin.org) which lists, among other items, more than 150 national parent information organizations. All this at the immediate—and literal—fingertips. As the information base in child development and the information resources for parents increase geometrically, we have a concomitant geometric decline in the amount of time it takes to access that information, along with a geometric decline in the amount of time it takes for information to go from academic debate and the research laboratory to translation and mutation into advice books, into the Sunday supplement articles, onto the radio and television talk shows, to be formed exquisitely and unforgettably into the media sound bite. All this—child development made easy for the PITS, the person in the street—is an understandable response to expressed and unexpressed needs of parents, and caregivers, and teachers. The media are only responding to the market. And responsive they are, proffering advice made sometimes too attractive, especially if it is made up of one part fact to three or four parts exaggeration, hype, and overgeneralization. What we have is a seemingly insatiable hunger for simple answers to simple questions. How else can we explain the relatively frequent headlines that claim the single-variable responsibility for developmental outcomes: it’s all about peers (parents are irrelevant), or the genes—more specifically, a gene—for shyness, for intelligence, for personality, for grammar. “First Gene to Be Linked with High Intelligence Is Reported Found” headlined science writer Nicholas Wade’s (1998) article for The New York Times with the tantalizing inset teaser: “A new clue in the debate over what determines ability.” “Variant Gene Tied to a Love of New Thrills” was The New York Times headline for Natalie Angier’s (1996) rather informed article about the “partial genetic explanation for a personality trait called ‘novelty seeking.’” Even when the texts of such articles make reference to appropriate qualifications and
Child Development and the PITS 5 note the complexities, the headlines convey the simpler message. These simpler messages get tucked into minds and shape popularized ideas into present and future belief systems. A number of years ago it was bonding, with dire implications foretold if there was no mother-infant skin-to-skin contact in the first hours after birth. More recently, the popular media have reported new recommendations, liberally mixed with political ideology, about infant feeding on demand needing to give way to feeding on strict schedules as corrective for generations of poorly disciplined children. Tomorrow, next month, next year, it will be other variables—identified in isolation, heralded as all-important if not all-determining. And there will be no surcease in supplying the stories for the reporters and the headline writers by those who, for a variety of reasons, some sincere and informed, some ideological and self-serving, are more than willing to satisfy the craving for simple answers to simple questions. This is not to deny that the ultimate scientific ideal is nothing if not the embodiment of the search for the simplifying and unifying assumptions that will integrate disparate pieces of evidence to explain highly complex phenomena. For sure, given the current state of affairs, our developmental science has a long way to go before we might achieve such scientific elegance—if ever we will. Though one might think, looking at the expansion of our database on children and their development, that we are making significant advances toward an elegant integration of our vast database into overarching theory. Witness the growth of “manuals” and “handbooks” from one volume to two volumes to four volumes to four fatter volumes (Carmichael, 1954; Damon, 1998; Mussen, 1970, 1983), to say nothing of the growth of the program of our own biennial meeting over the years. We have, I believe, the possibility of making significant progress toward the goal of a theoretical integration of our vast and growing database, but not if we persist in some of the peculiar tendencies of our science wherein each new theoretical formulation, rather than being tested by how well it accommodates existing data, is used to delegitimize data generated in the context of a previous theoretical fixation. I say delegitimize rather than ignore in the Kuhnian (Kuhn, 1970) sense, because, unlike in other sciences, where the success of new theoretical formu-lations is judged by how much of the existing verified data can be accommodated by the new theory, in human behavioral development new theories seem to be judged as successful by the numbers of adherents who are eager to reject data and principles generated by existing or older theories. Thus American Piagetian research ignored or rejected the data and principles established in the behaviorist tradition; behaviorism dismissed Gesellian data as uninformative and excoriated Freudian derived psychodynamic data. Behaviorism’s data, demonstrating more or less efficacious strategies for learning, were dismissed as nonlearning because they appeared to not consider more generic matters of cognition. Behaviorists were severely criticized and caricatured quite dismissively because they seemingly failed to include in the learning process the role of the “active child” acting on the environment to foster his or her own development. I’ve lost count of how often stimulus-response formulations of learning were said to be completely invalid because the S-R approach viewed the child as an entirely passive receptacle. One got the impression that critics were willing to suggest that it mattered little to behaviorists whether their participants were alert or anesthetized. To turn the
Annual progress in child psychiatry and child development 2000–2001
6
tables, how often have behaviorists dismissed discussions of data that included difficultto-operationalize speculations and propositions that are, in some important ways, the stuff of the imaginative musings that give rise to scientific and theoretical advances? How often have they eschewed data analysis techniques as representing group fictions?
GROWING CONSENSUS? All said, however, I detect important progress and some growing consensus in recent years, if not yet widespread agreement in our science, that recognizes a need to embrace data from a variety of theoretical perspectives in the service of formulating more overarching developmental theories. To be sure, we may just be in an era of a new set of buzzwords and phrases—dynamic, nonlinear, systems, plasticity, life-course trajectories, bioecological, person-in-context, reciprocal influences, mediators, connectionism, and attractors. It may also be said that we seem to be in an era of enthusiasm for models. In 1983, the first volume of the Handbook of Child Psychology was entitled History, Theory, and Methods (Kessen, 1983); in 1998, the first volume of the Handbook is entitled Theoretical Models of Human Development (Lerner, 1998). The models include Overton’s Bio/Social-Cultural Action Matrix (Overton, 1998), Gottlieb’s systems view of psychobiological development (Gottlieb, Wahlsten, & Lickliter, 1992), Fischer and Bidell’s dynamic, domain-specific, skill structure developmental web model (Fischer & Bidell, 1998), and Thelen and Smith’s dynamical systems and modified epigenetic landscapes (Thelen & Smith, 1994). Encouragingly, the current academic jargon and models involve more acknowledgments of complexity than has been previously true, driven in large part by the complexity of the data, especially in relation to large cross-sectional and longitudinal data sets. Against the media popularity of single-variable stories, the science itself is moving inexorably toward greater and greater data-driven, integrative theoretical complexity. An exception to this is behavioral genetics. In contrast to the dynamic nonlinear interactive models full of reciprocity between and among levels and variables, behavioral genetics presents a relatively nondynamic linear additive model that tries to assign percentages of variance in behavior and development that can be attributed to genes. The enterprise rests on the assumption that genetic influence can be expressed as a value accounting for a portion of the variance in a nondynamic linear equation for predicting behavioral functioning, and, furthermore, that individual experiences of shared and nonshared environments can be assessed inferentially by the degree of biological relatedness of individuals without empirical observations of experience (Hoffman, 1991; Horowitz, 1993). Behavioral genetics involves a relatively simplistic approach when compared with the kinds of dynamic system theories currently being elaborated. Perhaps that is why, in the mode of wanting simple answers to simple questions, behavior genetic reports are so media-attracting. However, so as not to seem to be repeating the practice I’ve just criticized of dismissing data in the face of new theoretical formulations, it needs to be said that the data reported in behavioral genetics studies involving degrees of
Child Development and the PITS 7 relationships among twins, siblings, and biologically unrelated individuals are in themselves interesting, even if it is doubtful that these relationships tell us anything about the direct and unmediated impact of genes. In formulating the more recent complex models of development one sees increasing skepticism about what is to be learned from assigning variance percentages to genes (e.g., Elman et al., 1998; Kagan, 1998). The skepticism is informed by approaches that see genes, the central nervous system, and other biological functions and variables as contributors to reciprocal, dynamic processes which can only be fully understood in relation to sociocultural environmental contexts. It is a perspective that is influenced by the impressive recent methodological and substantive advances in the neurosciences. Data from studies that employ neuroimaging techniques are providing extremely important information about structural plasticity in neuropsychological function. Most critically, this structural and functional plasticity across developmental time is being tied directly to the amplifications and constraints of the social/cultural contexts that determine the opportunities that children and adults have to experience and learn (Elman et al., 1998; Lewontin, Rose, & Kamin, 1984; Nelson & Bloom, 1997).
TOWARD AN INTEGRATIVE THEORY Let me suggest that these advances lead us, if not anywhere near the brink of an integrative theory and the elegance to be achieved by a set of unifying and simplifying assumptions, then at least toward a better understanding of the complex and dynamic nature of the relationships that impact development and the operation of developmental processes. Permit me to enter, not a new model of development per se, but a graphic to represent the range and complexity of what we must understand to achieve a fuller description of development and developmental processes (see Figure 1.1). It represents a way of thinking that I believe will accommodate and perhaps elaborate a number of the developmental models now being described and the data they are generating. In other words, this is not a de novo entrant into the arena of models but an attempt at a synthesis that might better organize our data and how we think theoretically. You will recognize in Figure 1.1 shades of a number of models and graphics by others with respect to organism-environment reciprocity (e.g., Gottlieb, Wahlsten, & Lickliter, 1998; Wachs, 1992) and efforts to parse the environment (e.g., Bronfenbrenner, 1979; Bronfenbrenner & Ceci, 1994; Horowitz, 1987; Horowitz & Haritos, 1998). In this model, as in some of the others, the assumption is made (supported by data) that from the moment of conception development is influenced by constitutional, social, economic, and cultural factors and that these factors, furthermore, continue in linear and nonlinear relationships, to affect development across the life span, with development broadly defined to accommodate both the increase and decrease in ability and function. Throughout the model, I use the word “experience” rather than “environment” to emphasize that the operative aspect of environment is experience. What is suggested by large amounts of data across many different studies (and not surprisingly to many in this audience) is that, taken together or in various linear and nonlinear combinations and
Annual progress in child psychiatry and child development 2000–2001
8
Figure 1.1 A depiction of the constitutional, social, cultural, and economic sources of influence on development with respect to the nature of experience and in relation to the circumstances of advantage, risk, and promise. permutations, constitutional, economic, social, and cultural factors provide the set of circumstances, or context, for development. These circumstances may, in aggregate, generally provide normal advantage, poor advantage, or high advantage. Unaggregated, as will be illustrated in a bit, they can also provide advantage or disadvantage in a particular developmental domain. In this schematic, the greater the presence of poorly advantaging circumstances, the more overall development is put at risk; the greater the presence of highly advantaging circumstances, the more promise for overall development. The circumstances that condition the possibilities of risk and promise begin with conception; past the moment of conception, in addition to the normal genetic and biological processes during the prenatal period, social, economic, and cultural variables of environmental origin, mediated by maternal biology, begin to operate. They contribute to setting the base of the child’s initial constitutional circumstances at birth. The point being made here is that already in the prenatal period, as a number of investigators have
Child Development and the PITS 9 shown, we have to consider experiential aspects of environmental origin, albeit mediated through maternal biology. Past the prenatal period, it becomes important and, I believe, useful to think about how to organize our thinking and data with regard to parsing the functional dimensions of experience in terms of what is the minimal level, amount, or nature of experience necessary for the development of the universal human behavioral repertoire—experience that is highly probable for the normally developing human organism; experience insured by the extensive amount of naturally occurring redundancy. Beyond the minimal level, I believe the data suggest there is a normal, highly likely range of experience provided postnatally for most children growing up in normal and near-normal environments. These experiences serve to sculpt and elaborate the basic species-typical universal human behaviors. They begin also to shape the vast repertoire of nonuniversal behaviors important to functioning in different social, cultural, and economic societies. The conundrum for many is to explain the regularities of the postnatal emergence of the normal universal species-typical behaviors in each individual child despite the seeming variations in the gross nature of environments. The nativist answer is recourse to instincts, to predetermined, architecturally and genetically driven explanations, both for the species as a whole and for the individuals in particular (Chomsky, 1965; Pinker, 1994; Spelke, Breinlinger, Macomber, & Jacobson, 1992; Spelke & Newport, 1998). To the Person in the Street these explanations seem to provide the simple answers to simple questions though the nativist position is by no means simplistic and the position is often supported by very interesting data. The alternative view and, I believe, the more compelling view is to consider that within all the gross environmental variations there is present the essential minimal experience necessary for the acquisition—the learning—of the basic universal behaviors of our species. There is a growing agreement that universal behaviors and physical structures are not built into the organism but that humans are, at the very least, evolutionarily primed to take advantage of the transactional opportunities provided by what Brandstäder (1998) sees as the universal physical and social ecologies available to all normal human organisms—the kinds of transactional opportunities so beautifully analyzed by Thelen and her colleagues with respect to early motor development (Thelen & Ulrich, 1991). As a result of these transactional experiences, the forms and functions of the universal developmental domains are constructed, whether as described in Thelen’s dynamic systems approach to motor development (Thelen & Smith, 1994; Thelen & Ulrich, 1991), or in Katherine Nelson’s (1996) powerful analysis and synthesis of the role of language in cognitive development, or in Kurt Fischer’s notion of the “constructive web” and his attempts to document the linear and nonlinear mechanisms involved in the construction and development of the hierarchies of skills (Fischer, 1980; Fischer & Bidell, 1998). These points of view are gaining in credibility because, with the aid of neuroimaging techniques (Nelson & Bloom, 1997), we are learning how actively responsive is the developing brain to experience. In all, the evidence is accumulating that the regularities of development are constructed as a result of the transaction of the individual with the seemingly big, buzzing, confusing, noisy environmental surround—an environmental context that provides a high level of redundant experiential opportunities for these universal capacities to be sculpted and, at the same time, for the variations across
Annual progress in child psychiatry and child development 2000–2001
10
environments to begin to shape the development of the nonuniversal behaviors that define individuals in linguistic, social, cognitive, economic, and cultural contexts (Horowitz & Haritos, 1998). For example, the capacity for language is a universal species-typical behavior of all normal humans. Its initial development and expres-sion rest on the normally occurring prenatal environment and the minimal level of the postnatal essential experience of hearing language and experiencing it in a social context. The acquisition of language is then further sculpted by the normal range of experiences involving the language of the cultural surround—Mandarin Chinese for one, Hebrew for another, Portuguese for another, and so on. I use the word “sculpted” here not to refer to some passive organism on which experience is writing the script but rather to an active collaboration of organismic (read constitutional) characteristics with experiential opportunities that impact the development of nonuniversal behaviors—nonuniversal behaviors that are determined in a social, cultural, economic, and constitutional context. In the normal range of experience, the capacity for language and the acquisition of a specific language is embedded in the social contexts that influence the use of language in communication, determining how language comes to serve the behavioral repertoire of social and cultural exchange expected of individuals in that cultural and social context. In turn, these experiences affect the development of constitutional characteristics in terms of brain structure and function with the constitutional characteristics also in dynamic relationship with experience and with the social, economic, and cultural contexts in which development is occurring. Until now, our attempts to parse and categorize experience have been relatively crude, crude as in Figure 1—suggesting, without much specification, that there is a minimal level of experience necessary for the development of basic universal behaviors, that a normal range of experience further enables the development of universal behaviors as well as the initial shaping of the nonuniversal behavioral repertoires. Beyond this, environments can provide for a range of normal additional experience and, further, extraordinary additional experience (all yet to be defined in terms of components and dynamic processes) which may or may not be the same across different environments. But there is a growing body of evidence that demonstrates the powerful effect of variations in experience, assuming some minima, on language development, on cognitive development, and on intelligence. In a detailed and painstaking study of the language input experiences and of the consequent language output of very young children growing up in different socioeconomic environments, Hart and Risley (1995) have shown that although all of the children they observed learned to talk and acquired the basic grammatical structure of English, children reared by professional parents had five times more words addressed to them over the first three years of life than did children reared by parents in poverty, with the concomitant effect of an increasingly widening gap between the recorded size of the children’s vocabulary so that by 3 years of age children reared by the more language-restricted parents in poverty had a vocabulary of less than 500 words, while those reared by language-rich professional parents had a vocabulary of about 1,100 words; children reared by middleand lower-income parents had a vocabulary of about 700 words. Huttenlocher and her colleagues (Huttenlocher, Levine, & Vevea, 1998) have shown the sensitivity of cognitive growth involving language, spatial operations, and concept
Child Development and the PITS 11 development to the experience reflected in the simple measure of amount of time spent in school. Brooks-Gunn, Klebanov, and Duncan (1996) have provided impressive evidence of the powerful impact of impoverished family resources on IQ such that when they controlled for the constellation of the social, economic, and cultural dimensions of poverty, the oft-reported black-white differences in IQ all but disappeared. It is almost 30 years since Sameroff and Chandler (1975), in their seminal chapter on the “continuum of caretaking casualty,” alerted us to the effects of the advantaging and disadvantaging macrosocial characteristics of environments on the postnatal developmental journeys of high-risk infants. The accumulating data since the 1970s has permitted us to refine our understanding of the variables and dynamics that impact the developmental outcomes of those infants. The data do, I believe, also permit us to conceptualize about the circumstances— constitutional, social, cultural, and economic—that conspire, effectively, to bestow normal, low, or high degrees of advantage during development—in general or with respect to particular developmental domains. The specific studies I have cited illustrate in a most general way that poorly advantaged environments, defined as providing children with impoverished or limited or sometimes only a little experience beyond the minimum, put the fullest realization of children’s development at risk by offering few or fewer opportunities for enriching additional experience or extraordinary additional experience. Conversely, highly advantaged environments, defined as providing many more opportunities for additional and enriched experiences, hold promise for the fullest realization of children’s development. At the extremes, at the ends of the continuum of advantage, a confluence of constitutional, social, economic, and cultural circumstances for poor advantage or enriched advantage can coalesce into what I call “swamping conditions.” That is, at the extremes a dense concentration of resources made possible, for example, by high socioeconomic advantage can have the effect of swamping development in a positive manner. Conversely, a dense concentration of disadvantaged circumstances can swamp development negatively. However, the picture is likely more complex. Swamping conditions at the extremes of disadvantage or advantage may or may not affect all domains of development, and they may have their origin in particular social, economic, cultural, or constitutional circumstances. For example, cerebral palsy or Down syndrome are constitutionally swamping conditions. Cerebral palsy is a swamping condition that involves severe constitutional compromises with respect to motor development. The presence of cerebral palsy may or may not have constitutional disadvantages in other developmental domains, and the child with cerebral palsy may be born into social, economic, and cultural circumstances that hold normal, low, or high degrees of advantage. In the case of cerebral palsy, its presence can render ineffective the minimal level of experience necessary for the development of the basic species behavioral universals related to human motor development. One can speculate that someday it will be possible, as it is now possible with the inborn metabolic disorder involved in phenylketonuria, to detect and then provide a physical/ biological/socially mediated intervention either prenatally or postnatally that would nullify cerebral palsy as a swamping condition for
Annual progress in child psychiatry and child development 2000–2001
12
normal motor development. In the meantime, this swamping condition may be ameliorated when children with cerebral palsy are provided with extraordinary additional experience designed to moderate the effect of the condition on motor functioning. This is not the occasion to explore the combinations and permutations and the linear and nonlinear functions that need to be taken into account in a refined analysis of the constitutional, social, economic, and cultural circumstances interacting with various degrees of experience, by domain and across time. Suffice it to say it is likely that the dynamics and constituents of developmental processes are not static across time, nor are they linear. Further, a systems analysis of these variables accommodates the idea that we are dealing also with the interactive impact of individual differences as well as the power of suddenly appearing or enduring variables to change the dynamics of the system, perhaps to function as disadvantaging swamping conditions: psychological trauma, cultural upheaval, physical disability and disease, and social chaos. In the same way, conditions of economic stability and affluence, social cohesion, high-quality education, and consistent and saturating extraordinary additional experience can function as advantaging variables and, if intensive enough, as advantaging swamping circumstances. We must recognize, too, the confluence of organism and environment or of particular constitutional and/or social and cultural circumstances that make for individual resilience in the face of adversity, and individual vulnerability in the face of advantage. As has been noted, poverty in our society is clearly a disadvantaging economic variable, although under certain constitutional social, historical, and cultural contexts its disadvantaging effects may well be attenuated (Elder, 1999; Werner & Smith, 1982, 1992), especially when not compounded by the added negative factors of racism and discrimination. Affluence is an obviously advantaging condition, although under certain constitutional social, cultural, and historical circumstances its advantaging power may be diminished. In other words, the degree to which any constitutional, social, economic, and cultural circumstance is relatively advantaging or disadvantaging is highly contextualized. Further, the functional consequences of these circumstances will rest strongly on the nature and extent of focused and fortuitous environmentally organized and mediated experiences. At the extremes, in certain domains the constitutionally or economically swamping conditions may well play stronger roles than social and cultural variables in determining the degree of advantage. The presence of cerebral palsy is a disadvantaging condition for motor development, as is Down syn-drome for mental development, but not necessarily for all aspects of social development. In the case of Down syndrome, we know that providing early extraordinary additional experience attenuates some of the mental retardation (Carr, 1992). In addition, children who may be constitutionally or otherwise advantaged with respect to extraordinary giftedness and talent in athletics, in music, in art, in language, typically require extraordinary additional experiences in learning, training, and opportunity for such gifts to be fully expressed and realized (Feldman, 1986). Toward an integrative theory of human behavioral development, the challenge for the approach outlined here, or for any such attempt, is to determine how well this kind of a theoretical approach accommodates, explains, and encompasses our reliable database. I believe we may now be nearer to some partially successful efforts in this regard than we
Child Development and the PITS 13 have been in the whole history of our discipline. That is reason to step back and acknowledge that as a result of the collective of our scientific enterprise across the globe, we can say, with some satisfaction, that we are indeed making important progress.
ON SOCIAL RESPONSIBILITY Of course, for the Person in the Street, our progress may not be all that comforting because it doesn’t lend itself to providing simple answers to simple questions. Yet it is often the simple answer that is wanted, the simple variable, the blanket relief from parental responsibility, or the blanket prescription that will fix what is wrong, or, prospectively, the blanket formula that will insure the best developmental outcome. Thus the popularity of the 75 bestsellers giving advice on how to raise the spirited, the strongwilled, the emotionally intelligent, the nonconforming, the happy child, to say nothing of how to increase your child’s IQ. Thus the popular media interest in conceptualizations that say not much will make a difference, just good enough parenting is all that is wanted. It is interesting to think that while “good enough parenting” (Scarr, 1992) may have some appeal, the idea of “good enough teaching” is currently quite out of sync with our expectations of schooling and the almost epidemic fervor in this country about raising academic standards and increasing the level of school achievement. Just a little inconsistency here, especially when, increasingly, at both the micro and macro levels, we are coming to understand parenting as teaching, the kind of nondidactic teaching embedded in the subtle and notso-subtle variations in children’s parentally organized experiences, the kind of parental teaching that increasingly appears to be critical for the developing child, especially in relation to the nonuniversal behavioral repertoire. Yet consider that if you give credence to the notion of “good enough parenting” and combine that with the popularized simple answer that it is really the genotype that is the determining factor and that little the parent does will make a difference, and if you assume that what is true for parental efforts holds true for the teacher in the classroom, then you have a seemingly scientific rationale for the failure to educate, a rationale you can claim is sanctioned by scientific authority citing specific facts. But unlike Gertrude Stein’s rose, a developmental fact is only a fact in a theoretical context, a lesson we should have learned well from Piaget, an understanding generally resisted by doctrinaire behaviorists. Keeping control of facts in relation to theoretical context becomes increasingly important as knowledge grows but also as the posing of simple questions and the desire for simple answers just does not abate. The urge to simplify and especially to geneticize is a strong one. I recall a request to reprint Figure 1 used in my book on developmental theories. I had labeled one of the dimensions in the figure as organismic and the other as environmental (Horowitz, 1987), but the colleague requesting to reprint the figure in a book had crossed out the word organismic and substituted the word genetic. No, I said, the two were not equivalent and, unless the original label was to be used, my permission would not be granted. Similarly, in this discussion, “constitutional” is not equivalent to “genetic,” and purposely so. Constitutional includes the expressed functions of genes—which, in
Annual progress in child psychiatry and child development 2000–2001
14
themselves require some environmental input—but constitutional includes the operations of the central nervous system and all the biological and environmental experiences that impact organismic functioning and make constitutional variables part of the dynamic and reciprocal interactions that change across the life span as they affect the development of and the decline of behavior. In this perspective, the scientific challenges before us are severalfold. One is, as I have already indicated, to make significant progress in identifying the functional units and roles of experience. We need to learn how best to parse experience for the purpose of seeing its role within the dynamic systems responsible for development. Another challenge is to integrate more fully into our account of behavioral development the evidence emerging from the neurosciences about the effect of experience in shaping neurological function and structure. Still another is to remain vigilant in submitting any new theoretical formulation to the test of how well it accommodates the reliable database of the phenomena it purports to cover. Beyond the scientific challenge, however, is the challenge of helping the Person in the Street to learn to ask less simple questions and the challenge of communicating our knowledge and making clear the limitations of our knowledge in the most socially responsible manner possible. A fact is a fact is a fact is not analogous to Gertrude Stein’s rose. Moreover, the image of Stein’s unyielding rose does not carry with it serious social implications for the fabric of a society, even though Stein’s formulation may have had some existential import and influence on asthetic appreciation and theory. The social import of our facts and their interpretation is something we must care about. For good or for ill, our knowledge base is of enormous interest to the Person in the Street. None of us can singlehandedly deter the determined maker of the sound bite but we can make it difficult. None of us can singlehandedly cause the quest for simple answers to disappear but we can consciously attempt to suggest, in every venue, in every forum, that at the present state of our discipline most simple questions about human behavior and development require complex, often incomplete and unsatisfying answers. If we accept as a challenge the need to act with social responsibility then we must make sure that we do not use single-variable words like genes or the notion of innate in such a determinative manner as to give the impression that they constitute the simple answers to the simple questions asked by the Person in the Street lest we contribute to belief systems that will inform social policies that seek to limit experience and opportunity and, ultimately, development, especially when compounded by racism and poorly advantaged circumstances. Or, as Elman and Bates and their colleagues said in the concluding section of their book Rethinking Innateness (Elman et al., 1998), “If our careless, underspecified choice of words inadvertently does damage to future generations of children, we cannot turn with innocent outrage to the judge and say ‘But your Honor, I didn’t realize the word was loaded.’” As SRCD has so clearly acknowledged in its effort to communicate responsibly what we know for the purpose of informing enlightened social policy, we must do so only if we repeatedly remind the people in the street who ask the simple questions that development is complex, that our theories are incomplete, and that we do not fully understand all the variables and systems in control of development and developmental processes, even though, I believe, we can now say that our growing database points to the
Child Development and the PITS 15 critical role of experience interacting with the organism in affecting the realization of human potential in all domains and across the life span.
ADDRESS AND AFFILIATION Corresponding author: Frances Degen Horowitz, The Graduate School and University Center, City University of New York, 365 Fifth Avenue, New York, NY 10016–4309; email:
[email protected]
REFERENCES Angier, N. (1996, January 21). Variant gene tied to love of new thrills. New York Times, p. A1. Brandtstäder, J. (1998). Action perspectives on human development. In R.M.Lerner (Ed.), W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 807–864). New York: Wiley. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U., & Ceci, S.J. (1994). Nature-nurture reconceptualized in developmental perspective: A bioecological model. Psychological Review, 101, 568– 586. Brooks-Gunn, J., Klebanov, P.K., & Duncan, G. (1996). Ethnic differences in children’s intelligence test scores: Role of economic deprivation, home environment, and maternal characteristics. Child Development, 67, 396–408. Cairns, R.B. (1983). The emergence of developmental psychology. In W.Kessen (Ed.), P.H.Mussen (Series Ed.), Handbook of Child Psychology: Vol. 1. History, theory, and methods (4th ed., pp. 41–102). New York: Wiley. Carmichael, L. (Ed.). (1954). Manual of child psychology. New York: Wiley. Carr, J. (1992). Longitudinal research in Down syndrome. In N.W.Bray (Ed.), International review of mental retardation research (Vol. 18, pp. 197–223). New York: Academic Press. Chomsky, N. (1965). Aspects of a theory of syntax. Cambridge, MA: MIT Press. Damon, W. (Ed.). (1998). Handbook of child psychology (5th ed.). New York: Wiley. Elder, G.H. (1999). Children of the great depression. Boulder, CO: Westview. Elman, J.L., Bates, E.A., Johnson, M.H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1998). Rethinking innateness. Cambridge, MA: MIT Press. Feldman, D. (1986). Nature’s gambit. New York: Basic Books. Fischer, K.W. (1980). A theory of cognitive development: The control and construction of hierarchies of skills. Psychological Review, 87, 477–531. Fischer, K.W., and Bidell, T.R. (1998). Dynamic development of psychological structures in action and thought. In R.M.Lerner (Ed.), W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 467–561). New York: Wiley.
Annual progress in child psychiatry and child development 2000–2001
16
Gottlieb, G., Wahlsten, D., & Lickliter, R. (1998). The significance of biology for human development: A developmental psychobiological systems view. In R.M. Lerner (Ed.), W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 233–274). New York: Wiley. Hart, B., & Risley, T.R. (1995). Meaningful differences. Baltimore: Paul H.Brookes. Hoffman, L.W. (1991). The influence of the family environment on personality: Accounting for sibling differences. Psychological Bulletin, 110, 187–203. Horowitz, F.D. (1987). Exploring developmental theories: Toward a structural/behavioral model of development. Hillsdale, NJ: Erlbaum. Horowitz, F.D. (1993). Bridging the gap between nature and nurture. A conceptually flawed issue and the need for a comprehensive new environmentalism. In R. Plomin & G.E.McClearn (Eds.), Nature, nurture & psychology (pp. 341–354). Washington, DC: APA Books. Horowitz, F.D., & Haritos, C. (1998). The organism and the environment: Implications for understanding mental retardation. In J.A.Burack, R.M.Hodapp, & E. Zigler (Eds.), Handbook of mental retardation and development (pp. 20–40). New York: Cambridge University Press. Huttenlocher, J., Levine, S., & Vevea, J. (1998). Environmental input and cognitive growth: A study using time-period comparisons. Child Development, 69, 1012–1029. Kagan, J. (1998). Three seductive ideas. Cambridge, MA: Harvard University Press. Kessen, W. (Ed.). (1983). In P.H.Mussen (Series Ed.), Handbook of child psychology: Vol. 1. History, theory and methods (4th ed.). New York: Wiley. Kuhn, T. (1970). The structure of scientific revolutions. Chicago: Chicago University Press. Kuo, Z.-Y. (1967). The dynamics of behavior development. New York: Random House. Lerner, R.M. (Ed.). (1998). In W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed.). New York: Wiley. Lewontin, R.C., Rose, S., & Kamin, L.J. (1984). Not in our genes. New York: Random House, Pantheon Books. Lieber, L.R. (1960). Mits, wits, and logic (3rd ed.). New York: Norton. Mussen, P.H. (Ed.). (1970). Carmichael’s manual of child psychology. New York: Wiley. Mussen, P.H. (Ed.). (1983). Handbook of child psychology (4th ed.). New York: Wiley. Nelson, C.A., & Bloom, F.E. (1997). Child development and neuroscience. Child Development, 68, 970–987. Nelson, K. (1996). Language in cognitive development: The emergence of the mediated mind. New York: Cambridge University Press. Overton, W.F. (1998). Developmental psychology: Philosophy, concepts and methodology. In R.M.Lerner (Ed.), W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 107–188). New York: Wiley. Phelps, J.A., Davis, J.O., & Schwartz, K.M. (1997). Nature, nurture, & twin research strategies. Current Directions in Psychological Science, 6, 117–121. Pinker, S. (1994). The language instinct: How the mind creates language. New York: William Morrow. Sameroff, A.J., & Chandler, M.J. (1975). Reproductive risk and the continuum of
Child Development and the PITS 17 caretaking casualty. In F.D.Horowitz (Ed.), Review of child development research (Vol. 4, pp. 187–244). Chicago: University of Chicago Press. Scarr, S. (1992). Developmental theories for the 1990s: Development and individual differences. Child Development, 63, 1–19. Sears, R.R. (1975). Your ancients revisited: A history of child development. In E.M. Hetherington (Ed.), Review of child development research (Vol. 5). Chicago: University of Chicago Press. Senn, M.J.E. (1975). Insights on the child development movement in the United States. Monographs of the Society for Research in Child Development, 40(3–4), Serial No. 161. Spelke, E.S., Breinlinger, K., Macomber, J., & Jacobson, K. (1992). Origins of knowledge. Psychological Review, 99, 605–632. Spelke, E.S., & Newport, E.L. (1998). Nativism, empiricism, and the development of knowledge. In R.M.Lerner (Ed.), W.Damon (Series Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 275–340). Spock, B. (1998). Dr. Spock’s baby and child care. New York: Pocket Books. Thelen, E., & Smith, L.B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press/Bradford Books. Thelen, E., & Ulrich, B.D. (1991). Hidden skills: A dynamical systems analysis of treadmill stepping during the first year. Monographs of the Society for Research in Child Development, 56(1), Serial No. 223. Wachs, T.D. (1992). The nature of nurture. Newbury Park, CA: Sage. Wade, N. (1998, May 14). First gene to be linked with high intelligence is reported found. The New York Times, p. A16. Werner, E., & Smith, R. (1982). Vulnerable but invincible: A study of resilient children. New York: McGraw-Hill. Werner, E.E., & Smith, R.S. (1992). Overcoming the odds. Ithaca, NY: Cornell University Press.
PART I: DEVELOPMENTAL ISSUES
2 Developing Mechanisms of Self-Regulation Michael I.Posner and Mary K.Rothbart
Child development involves both reactive and self-regulatory mechanisms that children develop in conjunction with social norms. A half-century of research has uncovered aspects of the physical basis of attentional networks that produce regulation, and has given us some knowledge of how the social environment may alter them. In this paper, we discuss six forms of developmental plasticity related to aspects of attention. We then focus on effortful or executive aspects of attention, reviewing research on temperamental individual differences and important pathways to normal and pathological development. Pathologies of development may arise when regulatory and reactive systems fail to reach the balance that allows for both self-expression and socially acceptable behavior. It remains a challenge for our society during the next millennium to obtain the information necessary to design systems that allow a successful balance to be realized by the largest possible number of children.
INTRODUCTION We believe that understanding self-regulation is the single most crucial goal for advancing an understanding of development and psychopathology. Early in this century, Freud (1920) argued that the ego and superego developed to regulate largely unconscious motivational systems. In the latter part of this century mechanisms of self-regulation have begun to be uncovered through the study of attention and effortful control. There is also substantial reason to believe that understanding mechanisms of self-regulation in normal individuals will lead to advances in diagnosis, prevention, and possibly treatment of developmental problems like attention deficit disorder and learning disabilities. In turn, studies of these mechanisms in the developmental disorders will enhance our understanding of normal functioning. Self-regulation involves complex questions about the nature of volition and its relation to our genetic endowment and social experience. Much of the work on self-regulation has been purely behavioral. This is true in both attention studies carried out within cognitive psychology and studies of effortful control as a temperamental dimension. The lack of appropriate methods to study the physiology of the human brain has previously led to an
Annual progress in child psychiatry and child development 2000–2001
20
understandable hesitation in thinking about these processes at the neurosystems level. Kandel (1998, 1999), however, has argued persuasively that new concepts in neuroscience now make it possible to attempt to relate higher level cognitive concepts to underlying brain systems. His goal is to use modern neuroscience to reinvigorate the psychoanalytic approach to the mind, and he stresses the role of unconscious early experience in shaping the brain systems that control adult behavior. Even if such connections prove to be as yet premature, there is little question that they will be major topics in the coming years. A major goal of our chapter is to help the reader understand how new developments related to neural plasticity and neuroimaging have transformed the potential for understanding mechanisms that provide voluntary control of brain systems. Although Kandel (1998, 1999) has emphasized the relation of genetic and cellular processes to psychoanalytic concepts and therapy, our chapter concentrates mainly at the neurosystems level and deals with the mechanisms that produce voluntary control of our thoughts and actions. Discoveries within neuroscience have moved the field toward viewing the brain as plastic and open to influence by experience (Garraghty, Churchill, & Banks, 1998; Merzenich & Jenkins, 1995). The advent of neuroimaging has provided new tools for testing hypotheses about how the brain changes with experience and for exploring the behavioral mechanisms of self-regulation (Posner & Raichle, 1994). In this chapter we first examine some of the historical background for considering attention networks as mechanisms of self-regulation in the human brain. Next, we take advantage of imaging methods to examine how the brain might be altered by experience on a time scale from milliseconds to years. We then examine the role of high-level attentional networks as a vehicle for self-regulation and consider evidence that similar brain areas control regulation of emotion and cognition. We consider how individuals differ in effortful control and what some of the consequences of those differences might be for normal and pathological development. In our final section, we speculate on future developments in this field.
HISTORY Within cognitive psychology, the mechanisms thought to be involved in self-control are collectively called attention. In 1958, Donald Broadbent summarized British work in the field of attention in his volume Perception and Communication. He proposed a filter that held back messages from an unattended channel to keep them from interfering with selected input. Broadbent’s beautiful studies, summarized in nearly every textbook in psychology, provided a basis for studying how we make a selection of relevant information from the masses of potential input. The studies reviewed in his book viewed attention as a high-level skill that allowed some experts to perform selective feats such as simultaneous translation and even novices to have a role in selecting their environment. There were challengers to Broadbent’s ideas, but it is remarkable, in view of the four decades that have passed since 1958, how even his strongest critics have followed his general ideas. For example, Anne Treisman (1969) showed that the filter could better be
Developing mechanisms of self-regulation 21 described as an attenuator, with much less interference when input was to separate modalities (eye and ear), rather than to one. Norman (1969) argued it would be better to see the filter as operating later in the system, after input had already activated material stored in long-term memory. Indeed, experiments rather quickly established that familiar words could look up their meanings even prior to being perceived (Posner, 1978). Allport (1980) challenged whether limited capacity was related to attention, and argued that interference instead resulted from contradicting behavioral task demands. However, all of the ideas about attention that dominated cognitive journals for the last half century were clearly derivatives of the basic question Broadbent (1958) had posed about selective listening: How was it that some aspects of the input were perceived and others not? The connection between selective processes in perception and more general issues of selfregulation had to await a link between cognitive and the neurophysiological level of analysis. An important early link between studies of attention within human cognition and those using the methods of neurophysiology was provided by Sokolov (1963) in his treatment of the orienting reflex. The orienting reflex provided a physical basis for filtering input and presaged the intense interest within neurophysiology in how attention might modulate activity within sensory specific areas (Hillyard & Lourdes, 1998). The concept of the orienting reflex was readily adapted to the study of preverbal infants who could not be instructed as to where to attend by experimenters (see review by Ruff & Rothbart, 1996). Ruff and Rothbart (1996) identify landmark periods in the first year of life with regard to orienting to objects and control of distress and in the second year and beyond in children’s ability to plan and regulate cognitive skills. It was a relatively easy step to identify these changes in the ability of the child to regulate behavior with the development of brain areas that carried out attentional selection in adults. This step, however, has a powerful consequence. It allows us to transfer knowledge on the anatomy and circuitry of attentional networks (Posner & Raichle, 1994) to the development of orienting and regulation in infants and adults (Posner & Rothbart, 1998), providing mechanisms for understanding self-regulation and its development. In the following, we consider changes in the human brain that might reflect both the rapid switch of content that occurs when adult subjects shift the focus of their attention and the much slower accumulation of the ability to control attention that occurs over the early years of development.
PLASTICITY In neuroscience, the issue of plasticity in brain activity has been discussed mainly at the synaptic level. For example, correlated neural firing among neurons in contact with each other leads to a change in the probability of one neuron being able to induce firing in the other. This principle of learning, first discussed by Hebb (1949), has been shown to be a basic principle for synaptic plasticity. The use of neuroimaging methods, however, has provided an altogether different level of analysis of plasticity. Instead of individual synapses, the focus is on the question of how experience influences the set of neural areas active within a task and their time
Annual progress in child psychiatry and child development 2000–2001
22
course of activation. This work has begun to allow us to consider possible neural mechanisms for many of the kinds of changes involved in children’s learning and education. Table 2.1 indicates some of the ways in which the person’s own activity or learning from external-based events might work to change brain circuitry on a temporary or more permanent basis.
TABLE 2.1 Mechanisms of Plasticity
Time 1. Milliseconds 2. Seconds to minutes 3. Minutes to days 4. Weeks 5. Weeks 6. Years
Phenomenon Shifts in attention Priming
Mechanism Amplification
Reference Corbetta et al. (1990)
Tuning
Jiang et al. (2000)
Practice New associations Rule learning Development
Pathway Connections
Raichle et al. (1994) McCandliss et al. (1997)
Structures Attention networks
McCandliss et al. (1997) Posner & Rothbart (1998)
The top row of Table 2.1 refers to the finding that attention allows rapid changes in neural activity in local brain areas. Neuroimaging methods are sensitive to changes in blood flow that accompany neural activity. When a brain area is being used to perform computations in high-level skills, it will increase in activity (Corbetta, Miezin, Dobmeyer, Shulman, & Petersen, 1990). As children learn a new skill, they may show a high level of variability as they try different strategies (Siegler, 1997). Each of these strategies is represented by a connected set of neural areas that carry out particular computations in some order. In adult studies, it is possible to demonstrate how a particular strategy may assume momentary dominance. Attention can provide priority to some computations, reprogramming the organization of the circuits by which tasks are executed. Priority is produced by amplifying the amount of neural activity within the area performing the computation. Often this is done voluntarily, as one tries to select a set of operations that seem most appropriate to a given task. This is what we call effortful control by attention. However, strategies may also arise from the physical situation. In the presence of a calculator, the person may enter numbers and press the appropriate key. If the calculator is absent, the numbers may be written down and the operations perform mentally. In this way, the environment primes one network of areas rather than another. Priming (row 2 of Table 2.1) is produced by the presentation of a sensory event (e.g., the calculator mentioned in the preceding), or by thought (e.g., the activation of a visual or auditory word), which changes the processing pathway so that stimuli sharing some or the entire pathway will be processed more efficiently. Priming can produce reduced reaction time for responding to a related target that follows the prime. Neuroimaging and cellular studies suggest that the number of neurons activated by a primed target is reduced over
Developing mechanisms of self-regulation 23 those activated in nonprimed target processing. The prime apparently tunes the neurons involved in the target event so that only those most appropriate to processing the subsequent target are activated (Ungerleider, Courtney, & Haxby, 1998). The mechanisms of rows 1 (attention shifts) and 2 (priming) of Table 2.1 provide two means to improve the processing of a target. The first method requires the person to attend to the computation. The involvement of attention sets up a network for processing the stimulus, but at the cost of making attention less available for handling other events. Priming, however, may occur when attention is now no longer involved in the process, leaving it free to deal with other items. Nevertheless, the network remains active for a period to make the processing of previously attended computations available. Priming may also occur without attention as the result of a sensory process. A pathway that has been tuned by a priming event does not require current attention and thus does not produce interference with ongoing activity (Posner, 1978). Effortful control through attention and automatic pathway activation apparently achieve the same behavioral results by quite different underlying mechanisms. Practice on a set of already learned but not recently rehearsed associations (row 3 of Table 2.1) shows that automaticity can completely change the pathway used to accomplish the task. In one study using PET (Raichle et al., 1994), people were required to generate a use for a read or heard noun (e.g., pound as a use for a hammer). When a new list of words was presented there was activity in the left frontal and posterior cortex, anterior cingulate, and right cerebellum. Activity in the anterior insula was reduced over what was found in simply reading the words aloud. A few minutes of practice at generating an associated use shifted activation so that the left frontal and posterior areas important in generating a new use dropped away and the anterior insula, strongly activated during reading aloud, increased. When generating a given word became automated with practice, the same circuit was used as when skilled readers read words aloud. There appeared to be one circuit associated with the thought needed to generate a familiar but unpracticed use, and another when the task was automated, as in reading aloud or generating again a just practiced association. The circuit used for thought includes attentional mechanisms involving effortful control, whereas an automated circuit does not involve attention. In the study cited in the preceding, people are dealing with already wellknown associations; for example, the association between hammer and pound. Even when they have not practiced them recently, connections between hammer and pound are available. However, it is often necessary to acquire entirely new associations, as in learning the words of a foreign language. This involves establishing new connections in the brain (row 4 of Table 2.1) and may require many weeks of practice. In one study of learning 40 lexical items in a new artificial language, it took 20 to 50 hours of practice before the words showed the same superiority in reaction time usually found for reading the native language (McCandliss, Posner, & Givon, 1997). Even more complex than learning a few new associations is developing a whole system to carry out an important linguistic function (row 5 of Table 2.1). Studies using PET with literate adults have shown that areas of the visual system of the brain become active when strings of letters are possible words in English, whether they have meaning or not (Petersen, Fox, Snyder, & Raichle, 1990). This area of the brain is not active for nonsense
Annual progress in child psychiatry and child development 2000–2001
24
strings like a series of consonants. It seems to represent English orthography and has been called the visual word form system (Petersen et al., 1990). This system appears to be a left posterior function that serves to group letters of a word automatically into a single chunk. No such unified chunk occurs for a string of consonants. This system appears to require some years to develop. Evidence suggests it is not present in 7-year-olds, even in those who know how to read, and can be found in 10-year-olds to a limited degree. Moreover, once this system is developed, it appears to be strongly resistant to change (Posner & McCandliss, in press). The final row of Table 2.1 refers to changes in brain structures that develop over the early life span of the person. We have in mind the several years apparently required to develop attentional networks. One form of attentional control deals with the selection of information by orienting to a sensory modality or location (e.g., eye movements or shifts of visual attention in vision). Orienting shows marked development in the first year of life (Ruff & Rothbart, 1996). In the visual system, early development includes improvements in acuity, control of fixation, ability to disengage, preference for novel objects and locations, and the control of emotional distress (Ruff & Rothbart, 1996). A second form of attention shows strong development in the second year of life and after (Posner & Rothbart, 1998). It provides the child with the necessary independence from his or her sensory world to develop an agenda of his or her own. The development of this system and its significance for the child are described further in the following.
EXECUTIVE CONTROL The central issue of this section is to describe an approach for examining executive function as a developmental process in early childhood. The goal is to provide an experimental means to link individual differences in self-regulatory behaviors developing in early childhood to the maturation of underlying neural systems. Norman and Shallice (1986) developed a model of adult attention very much in the spirit of the Broadbent approach. They argued that a supervisory attention system comes into play in adults in resolving conflict, correcting errors, and planning new actions. There now appears to be excellent data that the ability to resolve conflict undergoes development in early childhood. Adult studies using positron emission tomography (PET) have been consistent in showing activity in midline frontal areas during tasks that might be thought to involve executive attention (Bush et al., 1998; Posner & DiGirolamo, 1998). One such task we have already discussed is generating the use of a word. When blood flow due to reading words aloud is subtracted from blood flow in generating a use, there is a strong activation in the frontal midline along with language related areas of the left hemisphere and in connected areas in the cerebellum. Subsequent studies using high-density electrical recording have shown that midfrontal activity is detected very early, about 150 ms after input. This suggests that the first activity involves marshaling the cognitive effort needed to generate a use beyond that needed in the relatively effortless task of reading aloud. This view also agrees with PET studies showing that midfrontal activity may occur even before the task starts, when subjects know that a difficult task will occur (Murtha,
Developing mechanisms of self-regulation 25 Chertkow, Beauregard, Dixon, & Evans, 1996). The most frequently studied task found to activate the frontal midline has been the Stroop effect. In this task, subjects must respond to one dimension of a stimulus (usually the ink color), while ignoring another prepotent dimension (usually the color word name). A summary of the many results on Stroop effect conflict shows a remarkable convergence on areas of the frontal midline in the anterior cingulate gyrus (Bush et al., 1998). Because the anterior cingulate is the major outflow of the limbic system, however, it seems reasonable that its main function would be related to emotion, not cognition, and there is clear evidence that anterior cingulate activity is a part of the brain’s system for evaluating pain (Rainville, Duncan, Price, Carrier, & Bushness, 1997) and for distress vocalization (Devinsky, Morrell, & Vogt, 1995). The pain studies have shown cingulate activity when heat stimuli were judged as painful in comparison to merely warm. Moreover, the cingulate activity appears to be more related to the amount of subjective distress caused by the pain than to the intensity of the sensory stimuli involved (Rainville et al., 1997). When an effort was made to control the distress produced by a given stimulus using hypnotic suggestion, the amount of anterior cingulate activation reflected felt distress, whereas the somatosensory cortex reflected stimulus intensity. Recent studies of negative emotion in adults have suggested that distress is also related to activity in the amygdala (Davidson & Sutton, 1995). When pictures depicting frightening or horrible scenes are shown to subjects, there is strong activation of the amygdala, and evidence now exists that activation of the amygdala can be modulated by frontal activity (Davidson & Sutton, 1995). There is some evidence that cingulate activity is related to our awareness of emotion rather than to the emotion itself. To measure emotional awareness, people are asked to describe how they feel about situations. Their written responses are coded for use of emotional terms and descriptors (Lane & Schwartz, 1992) and the resultant score is taken as a measure of their emotional awareness. In a recent study, twelve subjects were shown each of three highly emotional movies and three neutral movies during a PET scan (Lane et al., 1996). Differences in anterior cingulate blood flow between the emotional and neutral movies were positively related to the person’s level of emotional awareness. These data suggest that something about awareness of emotions during sad or happy events is related to changes in the anterior cingulate. This result is similar to the finding discussed in the preceding indicating that cingulate activity is more related to the painful feelings than to the intensity of the stimulus inducing the pain (Rainville et al., 1997). Control of distress is a major task for the infant and caregiver in the early months of life, and attention plays an important role in this regulation (Harman, Rothbart & Posner, 1997). In the first few months, caregivers help control distress mainly by holding and rocking. Increasingly, in the early months, visual orienting is also used. Caregivers then attempt to involve the child in activities that will occupy their attention and reduce their distress. These interactions between infant and caregiver may train the infant in control of distress and lead to the development of the midfrontal area as a control system for negative emotion. Later, when similar cognitive challenges arise, a system for regulating remote brain areas may be already prepared. Many psychologists agree with Denckla (1996) that “the difference between the child
Annual progress in child psychiatry and child development 2000–2001
26
and adult resides in the unfolding of executive functions” (p. 264). Luria (1973) also referred to the development of a higher level voluntary social attention system. More voluntary attentional mechanisms and individual dif-ferences in executive attention have important implications for the early development of behavioral and emotional control (Rothbart & Bates, 1998). In an early example of cognitive control in a limited domain, Diamond (1991) showed the stages from 9 to 12 months in the child’s resolving conflict between reaching along the line of sight in order to retrieve an object in a box. At 9 months, the line of sight dominates completely. Even if the infant’s hand touches the toy through the open side of the box, if its movement is not in line with the side the child is looking at, the infant will withdraw the hand and reach along the line of sight, striking the closed side. Three months later, infants are able to look at a closed side but reach through the open end to retrieve the toy. However, being able to reach for a target away from the line of sight is only a very limited from of conflict resolution. Gerstadt, Hong, and Diamond (1994) studied verbal conflict modeled on the Stroop paradigm in children as young as 3.5 years. Two cards were prepared to suggest day and night to the children: one depicted a line drawing of the sun, the other a picture of the moon surrounded by stars. Children in the conflict condition were instructed to say day to the moon card and night to the sun card. Children in the control condition were divided into two groups and instructed to say day or night to either a checkerboard or ribbon card. At every age, accuracy scores were significantly lower for conflict relative to control trials. Other efforts have been made with Stroop-like tasks (Jerger, Martin, & Piozzolo, 1988) and with the Wisconsin card sort task (Zelazo, Reznick, & Pinon, 1995) to study children as young as 31 months; little evidence of successful inhibitory control below 3 years has been found. We believe that children as young as 18 months might be undergoing development in frontal midline areas that would allow the limited conflict resolution related to eye position to become more general. We had found that children at 18 months could show context-sensitive learning of sequences (Clohessy, Posner, & Rothbart, in press). This is a form of learning that, in adults, appears to require access to the kind of higher level attention needed to resolve conflict. Adults can learn sequences of spatial locations implicitly when each location is invariably associated with another location (e.g., locations 13241324). This occurs even when the adult is distracted with a secondary task known to occupy focal attention (Curran & Keele, 1993). The implicit form of skill learning seems to rely mainly upon subcortical structures. However, when distraction is present, adults are not able to learn context-sensitive sequences (e.g., locations 123213) in which each association is ambiguous. We found that infants as young as 4 months could learn the unambiguous associations, but not until 18 months did they begin to show the ability to learn ambiguous or context-sensitive associations (e.g., locations 1213). Individual children showed wide differences in their learning abilities, and we found that the ability to learn context-sensitive cues was positively related to the caregiver’s report of the child’s vocabulary development. According to the analysis of the last section, a more direct measure of the development of executive attention might be reflected in the ability to resolve conflict between simultaneous stimulus events as in the Stroop effect. Because children of this age do not
Developing mechanisms of self-regulation 27 read, we reasoned that the use of basic visual dimensions of location and identity might be the most appropriate way to study the early resolution of conflict. The variant of the Stroop effect we designed to be appropriate for ages as young as 2 to 3 years involved presenting a picture depicting a simple object on one side of a screen directly in front of the child and requiring the child to respond with a key that matched the stimulus they were shown (Gerardi-Caulton, in press). The appropriate key could be either on the side of the stimulus (compatible trial) or on the side opposite the stimulus (incompatible trial). The child’s prepotent response was to press the key on the side of the target irrespective of its identity; however, the task required the child to inhibit the prepotent response and to respond instead based on identity. The ability to resolve this conflict is measured by the accuracy and speed of their key-press responses. Results of the study strongly suggested that executive attention undergoes dramatic change during the third year of life. Performance by toddlers at the very beginning of this period was dominated by a tendency to repeat the previous response. Perseveration is associated with frontal dysfunction, and this finding is consistent with the idea that executive attention is still very immature at 24 months. Even at this young age, however, toddlers were already showing a significant accuracy difference favoring compatible over incompatible trials. By the second half of the third and beginning of the fourth year, children showed a strikingly different pattern of responses. Children now performed with high accuracy for both compatible and incompatible conditions, showing the expected slowing for incompatible relative to compatible trials. The developmental transition appeared to occur at about 30 months. It was also possible to examine the relationship of our laboratory measures of conflict resolution to children’s performance on a battery of tasks requiring the child to exercise inhibitory control over their behavior. We found substantial correlations between these two measures. Even more impressive, elements of the laboratory task were significantly related to aspects of temperamental effortful control and negative affect. Children who were less slowed by conflict were described as showing lower negative affect. As we have seen, cingulate activity would be expected to relate well at this age to control of distress. It appears that the cognitive measure of conflict resolution has a substantial relation to the aspects of the child’s self-control that parents can report.
INDIVIDUALITY Temperament refers to individual differences in motor and emotional reactivity and selfregulation (Rothbart & Bates, 1998). The temperamental vari-able related to the development of executive attention is called effortful control, representing the ability to inhibit a dominant response in order to perform a subdominant response. The construct of effortful control is extremely important in understanding the influence of temperament on behavior. Until recently, almost all of the major theories of temperament have focused on temperament’s more reactive aspects related to positive and negative affect, reward, punishment, and arousal to stimulation. Individuals were seen to be at the mercy of their dispositions to approach or avoid a situation or stimulus, given reward or punishment cues. More extraverted individuals were expected to be sensitive to reward and show
Annual progress in child psychiatry and child development 2000–2001
28
tendencies to rapid approach; more fearful or introverted individuals, sensitive to punishment, were expected to show inhibition or withdrawal from excitement (Gray, 1987). Systems of effortful control, however, allow the approach of situations in the face of immediate cues for punishment, and avoidance of situations in the face of immediate cues for reward. The programming of this effortful control is critical to socialization. The work of Kochanska (1995) indicates that the development of conscience is related to temperamental individual differences in effortful control. Kochanska and colleagues found significant prediction from infants’ 9-month sustained attention to their contemporaneous restraint in touching a prohibited toy and to a multitask behavioral battery assessing effortful control at 22 months (Kochanska, Murray, & Harlan, 1999; Kochanska, Tjebkes, & Forman, 1998). In two large longitudinal studies (32 to 66 months and 9 to 45 months), Kochanska and her colleagues assessed children’s effortful control in a laboratory test battery (Kochanska, Murray, & Coy, 1997; Kochanska, Murray, Jacques, Koening, & Vandegeest, 1996, 1999). Although the number and difficulty of tasks was varied to assure developmental appropriateness, beginning at age 30 months, children’s performance was highly consistent across tasks, suggesting that they all measured a common process that developed over time. Children were also remarkably stable in their performance across time, with the stability of a composite measure of effortful control approaching that of some of the most enduring of traits such as intelligence or aggression. In addition, children’s performance and their parents’ reports about their temperamental effortful control capacities in their daily lives also converged significantly. Other research suggests longer-term stability of executive attention during childhood. In Mischel’s work, for example, the number of seconds delayed by preschool children while waiting for physically present rewards predicted their parent-reported attentiveness, ability to concentrate, and control over negative affect when the children were adolescents (Mischel, 1983; Shoda, Mischel, & Peake, 1990). Although temperament researchers had originally believed that temperament systems would be in place very early in development and change little over time (e.g., Buss & Plomin, 1975), we have since learned that temperament systems follow a developmental course (Rothbart, 1989; Rothbart & Bates, 1998). Children’s reactive tendencies to experience and express negative and positive emotions and their responsivity to events in the environment can be observed very early in life, but children’s self-regulatory executive attention develops relatively late and continues to develop throughout the early school years. Because executive attention is involved in the regulation of emotions, some children will be lacking in controls of emotion and action that other children can demonstrate with ease. Questionnaire studies of 6- to 7-year-olds have found a broad effortful control factor to be defined in terms of scales measuring attentional focusing, inhibitory control, low intensity pleasure, and perceptual sensitivity (Rothbart, Ahadi, Hershey, & Fisher, 1997). Effortful control scores are negatively related to children’s scores on a negative affectivity factor. This negative relation is in keeping with the notion that attentional skill may help attenuate negative affect. An interesting example involves the negative relation between effortful control and aggression. Aggression relates negatively to effortful control and positively to surgency and negative affectivity, especially anger (Rothbart,
Developing mechanisms of self-regulation 29 Ahadi, & Hershey, 1994). Because effortful control makes no unique contribution to aggression, it may regulate aggression indirectly by controlling reactive tendencies underlying surgency and negative affectivity. For example, children high in effortful control may be able to direct attention away from the rewarding aspects of aggression, or to decrease the influence of negative affectivity by shifting attention away from the negative cues related to anger. Eisenberg and her colleagues, for example, found that 4to 6-year-old boys with good attentional control tend to deal with anger by using nonhostile verbal methods rather than overt aggressive methods (Eisenberg, Fabes, Nyman, Bernzweig, & Pinulas, 1994). Empathy is also strongly related to effortful control, with children high in effortful control showing greater empathy (Rothbart et al., 1994). In a study of elderly hospital volunteers, Eisenberg and Okun (1996) found attentional control to be positively related to sympathy and perspective taking, and negatively related to personal distress. In contrast, negative emotional intensity was positively related to sympathy and personal distress. Effortful control may support empathy by allowing the individual to attend to the thoughts and feelings of another without becoming overwhelmed by their own distress. Similarly, guilt or shame in 6- to 7-year-olds is positively related to effortful control and negative affectivity (Rothbart et al., 1994). Negative affectivity may contribute to guilt by providing the individual with strong internal cues of discomfort, thereby increasing the probability that the cause of these feelings is attributed to an internal rather than external cause (Dienstbier, 1984). Effortful control may contribute further by allowing the flexibility needed to relate these negative feelings of responsibility to one’s own specific actions and to negative consequences for another person (Derryberry & Reed, 1994, 1996). Consistent with these influences on empathy and guilt, effortful control also appears to play a role in the development of conscience. The internalization of moral principles appears to be facilitated in fearful preschool-aged children, especially when their mothers use gentle discipline (Kochanska, 1991, 1995). In addition, internalized control is facilitated in children high in effortful control (Kochanska et al., 1996). Here, we see the influence of two separable control systems, one reactive (fear) and one self-regulative (effortful control), regulating the development of conscience. Although fear may provide reactive inhibition and strong negative affect for association with moral principles, effortful control provides the attentional flexibility needed to link negative affect, action outcomes, and moral principles. These findings illustrate the importance of temperament in general and effortful control in particular to the child’s emotional, cognitive, and social development. These underlying temperament systems may also serve a central role in the self-organization of personality (Rothbart, Ahadi, & Evans, 2000). This is particularly evident in the functions of attention, which select and coordinate the most important information and contribute to the storage of this information in memory. Although much theorizing emphasizes children’s behavior and influences of the immediate environment, children think about their experiences and can use attention to “replay” their positive and negative experiences. We now revisit Table 1 in its relation to temperament and effortful control. Voluntary attention shifting may moderate the experience of negative affect (row 1), whereas
Annual progress in child psychiatry and child development 2000–2001
30
involuntary orientation to negative affect may limit attentional capacity. Further, automatic emotional priming (row 2) may influence the meaning of events to the child. Emotional reactivity will also be influential in developing a system of learning that allows us to interact well with others (row 4) and to learn more arbitrary rules connected with conscience (row 5). Finally, development of the attentional systems themselves will provide the capacities underlying the development of self-regulation (row 6). Across development, one would expect emotional and attentional processes to function together to progressively stabilize particular kinds of information, shaping the child’s representation of the self and world (Derryberry & Reed, 1994, 1996; Rothbart et al., 1994). The study of temperamental individual differences thus links effortful control mechanisms to issues of empathy, aggression, and conscience that represent central issues of child socialization, and leads us now to a discussion of development and psychopathology.
DEVELOPMENT AND PSYCHOPATHOLOGY The early part of this century saw the development of psychoanalysis (Freud, 1920). Beginning with the neurology known in his time, Freud uncovered unconscious mechanisms that code our implicit experience and proposed methods to prevent them from controlling the behavior of patients suffering from various forms of pathology. Since that time, advances in neuroscience have changed our ability to link specific brain mechanisms to behavior (Kandel, 1998, 1999). At the same time, progress in psychology has begun to specify the mechanisms whereby individuals regulate their feelings and thoughts. We now recognize that both automatic or unconscious impulses and conscious strategies work to control behavior. Future efforts should help us forge an understanding of these concepts at cellular and genetic levels (Albright, Jessell, Kandel, & Posner, 2000). The area of development and psychopathology examines the interplay between conscious and unconscious mechanisms in both normal and atypical persons (Cicchetti & Cohen, 1995). This approach to development stresses the diverse pathways by which early temperament is refined through experience (Cicchetti & Tucker, 1994). For example, children high in fear and low in selfevaluation may come to avoid achievement situations resulting in possible feelings of inadequacy, leading to even stronger fear or anxiety and avoidance in response to novel or challenging situations. This developmental progression, however, is not without recourse. Changes in the external or internal environment may lead to improvements in an individual’s ability to master developmental changes and thus to redirect a developmental trajectory. Neuroimaging should allow us an increasing ability to examine control mechanisms of the brain and to understand how their malfunction may form the basis for pathologies. For example, to display empathy to others requires that we interpret their signals of distress or pleasure. Imaging work in normal adults shows that sad faces activate the amygdala. As sadness increases, this activation is accompanied by activity in the anterior cingulate as part of the attention network (Blair, Morris, Frith, Perrett, & Dolan, 1999). It seems likely that the cingulate activity represents the basis for our attention to the distress
Developing mechanisms of self-regulation 31 of others. Psychopaths, for example, fail to show behavioral responses to sad faces and lack empathy to the distress of others. It seems likely that they would show either reduced activity in the amygdala or a loss of cingulate activation or both. If in one person the amygdala shows a strong response to the sadness of others, empathy would emerge quite naturally, but if little or no signal occurs, effortful control as influenced by socialization might still allow successful use of whatever signal was present. This is an example of how we might understand at a biological level the various pathways by which development produces either successful socialization or antisocial pathology. With the introduction of neuroimaging studies, it became possible to discover which brain areas became active during cognitive tasks. As discussed, many conflict tasks like the Stroop effect produced activation of the anterior cingulate (Bush et al., 1998). Washburn (1998) showed rhesus monkeys could be trained to perform a version of the Stroop effect known in humans to activate the cingulate. The monkeys showed many more errors on incompatible trials than do humans, however, despite many hundreds of trials at the task. It is as though the monkeys have somewhat less capacity for avoiding interference, despite very extensive training. A recent study was conducted with adults who suffer from attention deficit disorder. They performed conflict trials only slightly less efficiently than normals, but unlike normal controls they showed no evidence of anterior cingulate activation and instead showed greater activity on incompatible trials in the anterior insula (Bush et al., 1999). It is possible that the insula represents a more primitive pathway to output, one allowing for less effortful control (Raichle et al., 1994). Genetic studies of ADHD families have shown that they possess a mutation that affects the dopamine 4 receptor (LaHoste et al., 1996; Smalley et al., 1998). The dopamine 4 receptor is expressed in layer V of the cingulate. Although the anterior cingulate is an ancient structure, there is evidence that it has evolved significantly in primates. Humans and great apes appear to have a cell type found mainly in layer V of the anterior cingulate and the insula, which is not present in other primates (Nimchinsky et al., 1999). It is not known what the function of this cell is, but it appears to be a form of projection cell. There is evidence of development in the connectivity of this cell in childhood (Conel, 1959) and it is known that layer V of the cingulate expresses several dopamine receptors (Lidow, Wang, Cao, & Goldman-Rakic, 1998). Although there is as yet no direct evidence of the cellular basis of the cingulate activity found in neuroimaging studies, the importance of this area for emotional and cognitive tasks invites further exploration of linkage between the cellular architecture of this area and its function in self-regulation. Another disorder that produces a disruption of attentional control, as well as other emotional and cognitive problems, is schizophrenia. Benes (1999) has reported subtle abnormalities of the anterior cingulate in postmortem analyses of schizophrenic brains. She argues that the problem in schizophrenic brains may be a shift in dopamine regulation from pyramidal to nonpyramidal cells. She has also argued that these changes in the cingulate are related to circuitry involving the amygdala and hippocampus. These effects involve the D2 receptor and are strongest within layer II of the anterior cingulate. However, there are strong connections between layers II and V. The schizophrenia studies thus provide an entry to possible dysregulation of the anterior cingulate at a cellular level in a second abnormality noted for its attentional deficits.
Annual progress in child psychiatry and child development 2000–2001
32
FUTURE DIRECTIONS This paper has touched upon a number of areas in which we expect significant progress in the future. These include studies at the genetic, cellular, and synaptic levels, as well as at the neurosystems levels. In our understanding of temperamental individual differences, we also see a number of important di-rections for the next century. Findings suggest developmentally changing pathways through which early temperament, in conjunction with processes of socialization, can influence the development of social-cognitive processes. Tracing such developmental pathways to adaptive outcomes and the development of behavior problems and psychopathology will be among several promising directions for future work. Needed in the future is a more thorough understanding of the processes of temperament and how they develop, including surgency and extraversion, fear, and frustration, as well as attention. Many of these advances will come from affective and cognitive neuroscience, where to date much progress has been made in understanding the emotions (Davis, 1992; LeDoux, 1996; Panksepp, 1998) and attention (Posner & Raichle, 1994), as we have discussed in this paper. We expect progress in the study of brain structure and its underlying molecular genetics to be of great future importance. In addition, we will be looking toward improved temperament assessment methods that target multiple levels: behavior in laboratory paradigms, including marker tasks for the development of brain structures; behavior in naturalistic daily contexts; physiological measures; informant reports; and self-reports, including information about felt experience. Advances in assessment and empirical understanding will further allow us to link the past and future in the study of development. We envision moving toward bridging research on temperament in childhood with that on personality in adulthood (Caspi, 1998) within a coherent framework on individual differences that temperament can provide. The pathways between early temperament and future personality outcomes will, of necessity, be intricate, because child individuality unfolds in the context of social relationships, and continuity and change cannot be understood without considering the developmentally changing impact of social experience. To understand developmental pathways, we will need to disentangle complex interaction effects among early temperament predispositions, socialization processes, relationships, and culture. Recent research has begun to provide compelling support for such interactions (Bates, Pettit, Dodge, & Ridge, 1998; Belsky, 1997a, 1997b; Belsky, Hsieh, & Cranic, 1998; Kochanska, 1991, 1995, 1997; Nachmias, Gunnar, Mangelsdorf, Parritz, & Buss, 1996; Wachs & Gandour, 1983). Because of those complexities, we are likely to find examples of both equifinality and multifinality in development (Cicchetti, 1993). Temperamentally different children may arrive at similar or equivalent outcomes via different pathways. For example, in Kochanska’s (1995) studies, fearful toddlers whose parents used gentle discipline and fearless toddlers whose parents capitalized on positive motivation in a close relationship attained comparable levels of conscience. Temperamentally similar children’s
Developing mechanisms of self-regulation 33 developmental pathways may also diverge as a result of different effective experiences in relationships or varying cultural pressures (Rothbart, Ahadi, & Evans, 2000). Finally, in considering issues of developmental pathways, it remains the challenge of this new century, as it has for all previous periods of human history, to seek to balance the needs of the developing child in expressing his or her individuality with the needs of the society to regulate such expression. Whether by drugs, training, or social engineering, we can be sure that the struggle to control impulses will continue both within and among individuals. The case of attention deficit disorder is one interesting example. Empirical evidence favors the efficacy of drugs, which, like Ritalin, will provide many children with the help they need to attend to their school lessons (Swanson et al., 1998). However, it may also be possible that training of high-level attention during the periods when it is undergoing development would prevent the expression of the disorder, at least in some children. The use of drugs and training also does not rule out the idea that the social environment may itself be a serious cause of some pathology. Panksepp (1998), for example, suggests the importance of engineering the social environment of schools to provide access to play with peers that he feels is lacking in the current scene. His suggestion need not conflict with drug treatment and training, and, indeed, multiple avenues for change may be needed to provide the kind of human beings best able to thrive in the society we will have in the future. As students of normal and pathological development, we must continue attempts to understand the mechanisms involved in self-regulation so that we will have methods to help adapt our children to a changing environment. As members of society, we must try to use our knowledge in a way that enhances the number of children for whom a successful balance between self-expression and the demands of society can be found.
ACKNOWLEDGMENTS This work was supported by grants from the James S.McDonnell Foundation and Pew Memorial Trusts and by NIMH Grant 43361 to the University of Oregon and gratefully benefited from contributions by Grazyna Kochanska and Douglas Derryberry.
REFERENCES Allbright, T., Jessell, T., Kandel, E., & Posner, M.I. (2000). Neuronal communication: A problem for the 21st Century. Cell, 100, 51–55. Allport, D.A. (1980). Attention and performance. In G.Claxton (Ed.). Cognitive psychology: New directions (pp. 17–42). London: Routledge & Kegan Paul. Bates, J.E., Pettit, G.S., Dodge, K.A., & Ridge, B. (1998). Interaction of temperamental resistance to control and restrictive parenting in the development of externalizing behavior. Developmental Psychology, 34, 982–995. Belsky, J. (1997a). Theory testing, effect-size evaluation, and differential susceptibility to rearing influence: The case of mothering and attachment. Child Development, 68, 598–
Annual progress in child psychiatry and child development 2000–2001
34
600. Belsky, J. (1997b). Variation in susceptibility to environmental influence: An evolutionary argument. Psychological Inquiry, 8, 182–186. Belsky, J., Hsieh, K.-H., & Crnic, K. (1998). Mothering, fathering, and infant negativity as antecedents of boys’ externalizing problems and inhibition at age 3 years: Differential susceptibility to rearing experience? Development and Psychopathology, 10, 301–319. Benes, F. (1999). Model generation and testing to probe neural circuitry in the cingulate cortex of postmortem schizophrenic brains. Schizophrenia Bulletin, 24, 219–229. Broadbent, D.E. (1958). Perception and communication. London: Pergamon. Blair, R.J.R., Morris, J.S., Frith, C.D., Perrett, D.I., & Dolan, R.J. (1999). Dissociable neural responses to facial expression of sadness and anger. Brain, 1222, 883–893. Bush, G., Frazier, J.A., Rauch, S.L., Seidman, L.J., Whalen, P.J., Rosen, B.R., & Biederman, J. (1999). Anterior cingulate cortex dysfunction in attention-deficit/ hyperactivity disorder revealed by fMRI and the counting Stroop. Biological Psychiatry, 45, 1542–1552. Bush, G., Whalen, P.J., Rose, B.R., Jenike, M.A., McInerney, S.C., & Rauch, S.L. (1998). The counting Stroop: An interference task specialized for functional neuroimaging—validation study with functional MRI. Human Brain Mapping, 6, 270– 280. Buss, A.H., & Plomin, R. (1975). A temperament theory of personality development. New York: Wiley. Caspi, A. (1998). Personality development across the life course. In W.Damon & N. Eisenberg (Eds.), Handbook of child psychology: Vol. 3. Social, emotional and personality development (5th ed., pp. 311–388). New York: Wiley. Cicchetti, D. (1993). Developmental psychopathology: Reactions, reflections, projections. Developmental Review, 13, 471–502. Cicchetti, D., & Cohen D. (1995). Developmental psychopathology. New York: Wiley. Cicchetti, D., & Tucker, D.M. (1994). Development and self-regulatory structures of the mind. Development and Psychopathology, 6, 533–549. Clohessy, A.B., Posner, M.I., & Rothbart, M.K. (in press). Development of the functional visual field. Acta Psychologica. Conel, J.L. (1959). The cortex of the twenty-four month old infant (Vol. 6). Cambridge, MA: Harvard University Press. Corbetta, M., Miezin, F.M., Dobmeyer, S., Shulman, G.S., & Petersen, S.E. (1990). Attentional modulation of neural processing of shape, color, and velocity in humans. Science, 248, 1556–1559. Curran, T., & Keele, S.W. (1993). Attentional and nonattentional forms of sequence learning. Journal of Experimental Psychology: Learning, Memory, & Cognition, 19, 189–202. Davidson, R.J., & Sutton, S.K. (1995). Affective neuroscience: The emergence of a discipline. Current Opinion in Neurobiology, 5, 217–222. Davis, M. (1992). The role of the amygdala in fear and anxiety. Annual Review of Neuroscience, 15, 353–375. Denckla, M.B. (1996). A theory and model of executive function: A neuropsychological
Developing mechanisms of self-regulation 35 perspective. In G.R.Lyon & N.A.Krasnegor (Eds.), Attention, memory and executive function (pp. 263–278). Baltimore: Paul H.Brookes. Derryberry, D., & Reed, M.A. (1994). Temperament and the self-organization of personality. Development and Psychopathology, 6, 653–676. Derryberry, D., & Reed, M.A. (1996). Regulatory processes and the development of cognitive representations. Development and Psychopathology, 8, 215–234. Devinsky, O., Morrell, M.J., & Vogt, B.A. (1995). Contributions of anterior cingulate to behavior. Brain, 118, 279–306. Diamond, A. (1991). Neuropsychological insights into the meaning of object concept development. In S.Carey & R.Gelman (Eds.), The epigenesis of mind: Essays on biology and cognition (pp. 67–110). Hillsdale, NJ: Erlbaum. Dienstbier, R.A. (1984). The role of emotion in moral socialization. In C.E.Izard, J. Kagan, & R.B.Zajonc (Eds.), Emotions, cognition, and behavior (pp. 484–514). Cambridge: Cambridge University Press. Eisenberg, N., Fabes, R.A., Nyman, M., Bernzweig, J., & Pinulas, A. (1994). The relations of emotionality and regulation to children’s anger-related reactions. Child Development, 65, 109–128. Eisenberg, N., & Okun, M.A. (1996). The relations of dispositional regulation and emotionality to elders’ empathy-related responding and affect while volunteering. Journal of Personality, 64, 157–183. Freud, S. (1920). A general introduction to psychoanalysis. New York: Basic. Garraghty, P.E., Churchill, J.D., & Banks, M.K. (1998). Adult neural plasticity: Similarities between two paradigms. Current Directions in Psychological Science, 7 (3), 87–91. Gerardi-Caulton, G. (in press). Sensitivity to spatial conflict and the development of self regulation in children 24–36 months of age. Developmental Science. Gerstadt, C.L., Hong, Y.J., & Diamond, A. (1994). The relationship between cognition and action: Performance of children 3½–7 years old on a Stroop-like daynight test. Cognition, 53, 129–153. Gray, J.A. (1987). Perspectives on anxiety and impulsivity: A commentary. Journal of Research in Personality, 21, 493–509. Harman, C., Rothbart, M.K., & Posner, M.I. (1997). Distress and attention interactions in early infancy. Motivation and Emotion, 21, 27–43. Hebb, D.O. (1949). Organization of behavior. New York: Wiley. Hillyard, S.A., & Lourdes, A.V. (1998). Event-related potentials in the study of visual selection attention. Proceedings of the National Academy of Science USA, 9, 781–787. Jerger, S., Martin, R.C., & Pirozzolo, F.J. (1988). A developmental study of the auditory Stroop effect. Brain and Language, 35, 86–104. Jiang, Y., Haxby, J.V., Martin, A., Ungerleider, L.G., & Parasuraman, R. (2000). Complementary neural mechanisms for tracking items in human working memory. Science, 287, 643–646. Kandel, E.R. (1998). A new intellectual framework for psychiatry. American Journal of Psychiatry, 155, 457–469. Kandel, E.R. (1999). Biology and the future of psychoanalysis: A new intellectual framework for psychiatry revisited. American Journal of Psychiatry, 156, 504–524.
Annual progress in child psychiatry and child development 2000–2001
36
Kochanska, G. (1991). Socialization and temperament in the development of guilt and conscience. Child Development, 62, 1379–1392. Kochanska, G. (1995). Children’s temperament, mothers’ discipline, and security of attachment: Multiple pathways to emerging internalization. Child Development, 66, 597–615. Kochanska, G. (1997). Multiple pathways to conscience for children with different temperaments: From toddlerhood to age five. Developmental Psychology, 33, 228–240. Kochanska, G., Murray, K., & Coy, K.C. (1997). Inhibitory control as a contributor to conscience in childhood: From toddler to early school age. Child Development, 68, 263–277. Kochanska, G., Murray, K., & Harlan, E. (2000). Effortful control in early childhood: Continuity and change, antecedents, and implications for social development. Developmental Psychology, 36, 220–232. Kochanska, G., Murray, K., Jacques, T.Y., Koenig, A.L., & Vandegeest, K.A. (1996). Inhibitory control in young children and its role in emerging internalization. Child Development, 67, 490–507. Kochanska, G., Tjebkes, T.L., & Forman, D.R. (1998). Children’s emerging regulation of conduct: Restraint, compliance and internalization from infancy to the second year. Child Development, 69, 1378–1389. LaHoste, G.J., Swanson, J.M., Wigal, S.B., Glabe, C., Wigal, T., King, N., & Kennedy, J.L. (1996). Dopamine D4 receptor gene polymorphism is associated with attention deficit hyperactivity disorder. Molecular Psychiatry, 1, 121–124. Lane, R.D., Reiman, E., Ahern, G., Schartz, G., Davidson, R., Axelrod, B., & Yun, L. (1996). Anterior cingulate participates in the conscious experience of emotion. Psychosomatic Medicine, 58, 73. Lane, R.D., & Schwartz, G.E. (1992). Levels of emotional awareness: Implications for psychotherapeutic integration. Journal of Psychotherapy Integration, 2, 1–18. LeDoux, J.E. (1987). Emotion. In F.Plum (Ed.), Handbook of physiology: Vol. 5. Higher functions of the brain, Part 1 (pp. 419–460). Bethesda, MD: American Psychological Society. Lidow M.S., Wang, F., Cao, Y., & Goldman-Rakic, P.S. (1998). Layer V neurons bear the majority of mRNSs encoding the five distinct dopamine receptor subtypes in primary prefrontal cortex. Synapse, 28, 10–20. Luria, A.R. (1973). The working brain. New York: Basic Books. McCandliss, B.D., Posner, M.I., & Givon, T. (1997). Brain plasticity in learning visual words. Cognitive Psychology, 33, 88–110. Merzenich, M.M., & Jenkins, W.M. (1995). Cortical representation of learned behaviors. In P.Anderson, Ohvalby, O.Paulsen, & B.Hokfelt (Eds.), Memory concept (pp. 437– 451). Amsterdam: Elsevier. Mischel, W. (1983). Delay of gratification as process and as person variable in development. In D.Magnusson & V.P.Allen (Eds.), Human development: An interactional perspective (pp. 149–165). New York: Academic Press. Murtha, S., Chertkow, H., Beauregard, M., Dixon, R., & Evans, A. (1996). Anticipation causes increased blood flow to the anterior cingulate cortex. Human Brain Mapping, 4, 103–112.
Developing mechanisms of self-regulation 37 Nachmias, M., Gunnar, M., Mangelsdorf, S., Parritz, R.H., & Buss, K. (1996). Behavioral inhibition and stress reactivity: The moderating role of attachment security. Child Development, 67, 508–522. Nimchinsky, E.A., Gilissen, E., Allman, J.M., Perl, D.P., Erwin, J.M., & Hof, P.R. (1999). A neuronal morphologic type unique to humans and great apes. Proceedings of the National Academy of Science USA, 96, 5268–5273. Norman, D.A. (1969). Memory and attention. New York: Wiley. Norman, D.A., & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior. In R.J.Davidson, G.E.Schwartz, & D.Shapiro (Eds.), Consciousness and self-regulation (pp. 1–18). New York: Plenum Press. Panksepp, J. (1998). Attention deficit hyperactivity disorders, psychostimulants, and intolerance of childhood playfulness: A tragedy in the making? Current Directions in Psychological Science, 1, 91–98. Petersen, S.E., Fox, P.T., Snyder, A.Z., & Raichle, M.E. (1990). Activation of extrastriate and frontal cortical areas by visual words and word-like stimuli. Science, 249, 1041– 1044. Posner, M.I. (1978). Chronometric explorations of mind. Hillsdale, NJ: Erlbaum. Posner, M.I., & DiGirolamo, G.J. (1998). Conflict, target detection and cognitive control. In R.Parasuraman (Ed.), The attentive brain (pp. 401–423). Cambridge, MA: MIT Press. Posner, M.I., & McCandliss, B.D. (1999). Brain circuits during reading. In R.Klein & P.McMullen (Eds.), Converging methods for understanding reading and dyslexia (pp. 305–337). Cambridge, MA: MIT Press. Posner, M.I., & Raichle, M.E. (1994). Images of mind. New York: Scientific American Library. Posner, M.I., & Rothbart, M.K. (1998). Attention, self-regulation and consciousness. Philosophical Transactions of the Royal Society of London B, 353, 1915–1927. Raichle, M.E., Fiez, J.A., Videen, T.O., MacLeod, A.M.K., Pardo, J.V., Fox, P.T., & Petersen, S.E. (1994). Practice-related changes in human brain functional anatomy during nonmotor learning. Cerebral Cortex, 4, 8–26. Rainville, P., Duncan, G.H., Price, D.D., Carrier, B., & Bushness, M.C. (1997). Pain affect encoded in human anterior cingulate but not somatosensory cortex. Science, 277, 968–970. Rothbart, M.K. (1989). Temperament and development. In G.A.Kohnstamm, J.E. Bates, & M.K.Rothbart (Eds.), Temperament in childhood (pp. 187–247). New York: Wiley. Rothbart, M.K., Ahadi, S.A., & Evans, D.E. (2000). Temperament and personality: Origins and outcomes, Journal of Personality and Social Psychology, 78, 122–135. Rothbart, M.K., Ahadi, S.A., & Hershey, K.L. (1994). Temperament and social behavior in childhood. Merril—Palmer Quarterly, 40, 21–39. Rothbart, M.K., Ahadi, S.A., Hershey, K.L., & Fisher, P. (2000). Investigations of temperament at 3–7 years: The Children’s Behavior Questionnaire. Manuscript submitted for publication. Rothbart, M.K., & Bates, J.E. (1998). Temperament. In W.Damon & N.Eisenberg (Eds.), Handbook of child psychology: Vol. 3. Social, emotional and personality development (5th ed., pp. 105–176). New York: Wiley.
Annual progress in child psychiatry and child development 2000–2001
38
Ruff, H.A., & Rothbart, M.K. (1996). Attention in early development: Themes and variations. New York: Oxford University Press. Shoda, Y., Mischel, W., & Peake, P.K. (1990). Predicting adolescent cognitive and selfregulatory competencies from preschool delay of gratification: Identifying diagnostic conditions. Developmental Psychology, 26, 978–986. Siegler, R.S. (1997). Emerging minds: The process of change in children’s thinking. New York: Oxford University Press. Smalley, S.L., Bailey, J.N., Palmer, C.G., Cantwell, D.P., McGough, J.J., Del’Homme, M.A., Asarnow, J.R., Woodward, J.A., Ramsey, C., & Nelson, S.F. (1998). Evidence that the D4 receptor is a susceptibility gene in attention deficit hyperactivity disorder. Molecular Psychiatry, 3, 427–30. Sokolov, Y.N. (1963). Perception and the conditioned reflex. New York: Macmillan. Swanson, J.M., Sergeant, J.A., Taylor, E., Sonuga-Barke, E., Jensen, P.S., & Cantwell, D.P. (1998). Attention-deficit hyperactivity disorder and hyperkinetic disorder. Lancet, 35, 429–433. Treisman, A. (1969). Strategies and models of selective attention. Psychological Review, 76, 282–299. Ungerleider, L.G., Courtney, S.M., & Haxby, J.V. (1998). A neural system for visual working memory. Proceedings of the National Academy of Sciences USA, 95, 883– 890. Wachs, T.D., & Gandour, M. (1983). Temperament, environment, and six-months cognitive-intellectual development: A test of the organismic specificity hypothesis. International Journal of Behavioral Development, 6, 135–152. Washburn, D.A. (1994). Stroop-like effects for monkeys and humans: Processing speed or strength of association? Psychological Science, 5, 375–379. Zelazo, P.D., Resnick, J.S., & Pinon, D.E. (1995). Response control and the execution of verbal rules. Developmental Psychology, 31, 508–517.
PART I: DEVELOPMENTAL ISSUES
3 Implications of Attachment Theory for Developmental Psychopathogy L.Alan Sroufe, Elizabeth A.Carlson, Alissa K.Levy, and Byron Egeland
Bowlby’s attachment theory is a theory of psychopathology as well as a theory of normal development. It contains clear and specific propositions regarding the role of early experience in developmental psychopathology, the importance of ongoing context, and the nature of the developmental process underlying pathology. In particular, Bowlby argued that adaptation is always the joint product of developmental history and current circumstances (never either alone). Early experience does not cause later pathology in a linear way; yet, it has special significance due to the complex, systemic, transactional nature of development. Prior history is part of current context, playing a role in selection, engagement, and interpretation of subsequent experience and in the use of available environmental supports. Finally, except in very extreme cases, early anxious attachment is not viewed as psychopathology itself or as a direct cause of psychopathology but as an initiator of pathways probabilistically associated with later pathology.
INTRODUCTION From its inception, attachment theory was a theory of psychopathology as well as normal development. It was concerned both with the formation and normal course of attachment relationships and the implications of atypical patterns of attachment. Bowlby’s (1944) early consideration of 44 thieves revealed a consistent background of early parental privation in the lives of these young men. He reasoned that this was no mere coincidental association but, rather, had causal implications. Still, he doubted from the start that the connection was simple, direct, and linear—an environmental malignancy with an inevitable outcome. What, then, was the nature of this link? Attachment theory evolved in large part to answer this question. What Bowlby proposed was not just a theory of outcome, but a theory of process. In the attachment trilogy and other writings he presented a very specific set of propositions regarding the way in which early experience contributed to psychological health or pathology. He began in the first volume of the trilogy (1969) by clearly dissociating his
Annual progress in child psychiatry and child development 2000–2001
40
individual differences construct from causal trait constructs. Attachment referred to a pattern of organized behavior within a relationship, not a trait infants had in varying quantity (Sroufe & Fleeson, 1986; Sroufe & Waters, 1977). Attachment patterns were not immutable and not independent of subsequent experience. In Volume 2 (1973) he wrote that the development of the individual “turns at each and every stage of the journey on an interaction between the organism as it has developed up to that moment and the environment in which it then finds itself” (p. 364, italics added). Early experience frames, but also is transformed by, later experience. In brief, he proposed what would now be called a dynamic systems theory of psychopathology, based on a complex interaction of constituents over the course of development (e.g., Sameroff, 1989). Still, within this theory a special role for early experience in the initiation of important processes was specified. Thus, attachment theory provides a third alternative to the shadow box debate about whether early experience causes later outcomes in the manner of an immutable trait or bears only a coincidental association to outcome due to its link to later experience or third factors such as SES (Fox, 1997; Lamb, 1984; Lewis, 1997). The answer is, neither. Early experience often plays a critical role in the developmental dynamic that yields pathology, but this role is dependent on a surrounding context of sustaining environmental supports. At the same time, processes engendered in the context of early experience may influence the nature of later experience and the surrounding context itself. In this paper, core propositions from attachment theory regarding the complex relation between early experience and psychopathology will be elaborated. These include formulations concerning causation (and the interplay between the individual and context), the role of early experience, and the nature of early disturbance. This will be primarily a conceptual chapter, tracing the theoretical implications of the Bowlby/Ainsworth attachment theory for developmental psychopathology. Data presented will be for the purpose of illustrating these theoretical ideas, and will be drawn primarily from the Minnesota longitudinal parentchild study. Previously published data will not be presented in detail but will be accompanied by relevant citations. Many of these data and those of other researchers have been reviewed elsewhere (e.g., Weinfield, Sroufe, Egeland, & Carlson, in press). Data that have not been published before will be presented more completely, along with relevant statistical information. Likewise, we will be unable to review extensively topics such as representation and experiencedependent brain development, which are related to the issues pursued but are not central in this discussion. Such topics will be brought in where pertinent, and citations of recent comprehensive reviews will be provided.
CAUSE IN ATTACHMENT THEORY: ORGANIZATIONAL CONSTRUCT VERSUS TRAIT Cause is complex within attachment theory. In accord with ecological views such as those of Belsky (e.g., Belsky & Isabella, 1988) and Bronfenbrenner (1986), the child is seen as nested within a network of influences operating on many levels. Some contextual influences impact directly on the child, some indirectly through their impact on parenting.
Implications of attachment theory for developmental psychopathogy 41 Developmental context is emphasized, because “changes in circumstances can lead to changes in interaction and therefore to changes in relationships” (Vaughn, Waters, Egeland, & Sroufe, 1979, p. 974). At the same time, the child’s history of experience is a critical part of the developmental context. There is an ongoing transaction between the developing child and changing circumstances (cf. Sameroff & Chandler, 1975). The impact of current circumstances depends on the pattern of behavioral and emotional organization the child brings forward to that phase of development. Child and context are mutually transforming. Although we argue in the following that early experience has special significance, still it “…cannot be more important than later experience, and life in a changing environment should alter the quality of a child’s adaptation” (Sroufe, 1978, p. 56). The individual is the product of all of his or her experiences, not early experiences alone. Many of Bowlby’s ideas regarding the roles of prior experience and current circumstances in adaptation and in psychopathology are summarized within his concept of developmental pathways (Bowlby, 1969, 1973). Metaphorically captured by the dispersion and interconnection of tracks in a railway yard or branchings on a tree, this model embodies several key ideas (Sroufe, 1997). First, there are more lines or branches in the broad center of the larger array (there is great diversity in normality). Second, beginning on any major trunk allows a large number of possible outcomes due to subsequent branchings (multifinality). Thus, ongoing circumstances may support pursuance of potentially deviating developmental pathways or deflect the individual back toward more normal adaptation. Enjoining a pathway even early on does not determine final outcome but only initiates a set of possibilities. Cause is probabilistic, not deterministic. Third, the longer an outlying pathways is followed, the more unlikely becomes a return to centrality. In this theory psychopathology results from a successive series of adaptations. A pattern of anxious attachment in infancy may initiate such a process, but only if subsequent adaptations continue to represent deviation from positive functioning does psychopathology become likely. Change remains possible but, Bowlby (1973) argues, becomes quite difficult by adolescence if development has continued to go awry. Several specific, testable hypotheses may be derived from these initial formulations: (a) At any age, current quality of care will add to early attachment history in predicting pathology, given that adaptation is always the joint product of current circumstances and early history, (b) Likewise, broader aspects of current contexts, including relationships outside of the family and stresses and challenges of the period, also will increase prediction beyond early attachment; (c) a cumulative history of maladaptation will be more pathogenic than a single early period of poor functioning, with pathology ever more likely the longer a maladaptive pathway has been followed, and (d) change itself will be predictable in light of changes in stress and/or support. We and our colleagues have examined these hypotheses by carrying out a broad-based prospective, longitudinal study of 180 children, assessing quality of care in infancy, early childhood, and adolescence, along with assessments of relationships with peers and teachers, life adversities, and social support. Based on parent, self, and teacher behavior problem checklists at age sixteen (Achenbach & Edelbrock, 1986) and a clinical interview at age 17.5 years (the child form of the Schedule for Affective Disorders and
Annual progress in child psychiatry and child development 2000–2001
42
Schizophrenia; K-SADS) as the sources for psychopathology outcome, regression analyses were utilized to evaluate various combinations of variables. For example, to test the hypothesis regarding the significance of cumulative adaptational history, we compared the predictive power of infant attachment alone with the predictive power of attachment plus cumulative assessments of quality of care in the preschool years. Although anxious attachment in infancy by itself predicted a K-SADS psychopathology index (number and severity if past and present diagnoses; see Sroufe, 1997), those with a history of cumulative unsupportive care showed significantly more problems (13% vs. 15% of variance accounted for). In addition, we found that assessments of parent-child relationship problems at age 13 years (especially boundary dissolution and lack of emotional support or support for autonomy) and anxious attachment in infancy were more powerful than either alone, although each was significant separately (see Table 3.1). Likewise, ongoing assessments of peer relationship problems added to predictions of psychopathology based on attachment alone, R2 change (2,66)=.04, p <.01. Similar analyses have shown that attachment history and peer experiences in preschool or middle childhood together are better predictors of adolescent social competence than attachment alone (Englund, Levy, & Hyson, 1997; Sroufe, Egeland, & Carlson, in press).
TABLE 3.1 Hierarchical Regression Predicting Ratings of Psychopathology (K-SADS) at 17.5 Years from Avoidant and Disorganized Attachment in Infancy and Family Relationship Quality in Early Adolescence (N=120)
Step Independent Variables
R2
β
B
T
R2
Overall F df
Change Avoidant attachment .07 .27 .74 3.00** .07 8.99** 1,119 score (12–18 months) .18 .51 2.02* 2 Avoidant attachment .06 .25 .20 2.71** .13 8.41*** 2,118 .18 .18 2.04* score Disorganization rating (12–18 months) 3 Avoidant attachment .27 .27 3.03** score Disorganization rating Relationship rating (13 .05 .22 .21 2.59* .17 8.12*** 3,117 years) Note: Index of psychopathology in adolescence is 7-point rating of number and severity of KSADS diagnoses. *p<.05, **p<.01, ***p<.001. 1
Finally, in a number of papers we have evaluated the predictability of change. For example, in an early study we showed that quality of attachment itself changed in meaningful ways between twelve and eighteen months, given changes in parental life stress (Vaughn et al., 1979). Later, Erickson, Egeland, and Sroufe (1985) showed that those with histories of anxious attachment sometimes showed fewer behavior problems
Implications of attachment theory for developmental psychopathogy 43 than predicted in preschool. Such deflections in developmental course were best predicted by increased stability of social support for the primary caregiver. In summary, Bowlby did not say, nor does attachment-oriented research suggest, that early anxious attachment causes later pathology. He did claim that pathology would be a joint product of early experience and ongoing support or challenge, that cumulative maladaptation would be less easily changed than early anxious attachment, and change would be predictable. Each of these propositions has been amply supported by empirical research. In concluding this section, it is important to point out that the proposition regarding the power of cumulative adaptation does not mean that change cannot occur in adolescence or adulthood. Main, Kaplan, and Cassidy (1985) argued that the advent of formal operational thought may promote new opportunities for reflection upon, evaluation of, and integration of past experiences. Moreover, research with the Adult Attachment Interview (AAI) has uncovered numerous individuals who, despite indications of difficult life experiences, nonetheless have autonomous “states of mind” concerning attachment (so-called “earned secures”; Pearson, Cohn, Cowan, & Cowan, 1994; Phelps, Belsky, & Crnic, 1998). Clearly, some individuals have integrated very difficult experiences, and change can occur in the years of maturity. Such late-occurring changes are not counter to Bowlby’s idea that change becomes more difficult the longer the history of maladaptation. Positive changes may, in fact, not be likely in the face of continuous adversity and maladaptation from infancy through adolescence. When late changes occur, they may be built upon earlier foundations. Earned security on the AAI cannot really speak to this issue. First, the AAI was not designed to be a measure of early history, but rather as an assessment of the coherence of discourse concerning attachment (George, Kaplan, & Main, 1985; Main & Goldwyn, in press). Caution should be exercised in using it as a measure of history. There currently exists no empirical evidence as to the accuracy of reported early memories on the AAI. Second, even if veridical, which it seems likely to be to some extent, the AAI asks people to reflect on their memories from ages 5 to 12 years. At present, there is no evidence that adults have verbal access to experience from infancy. It is such preverbal experiences that Bowlby (1973) emphasized in his discussion of the enduring effects of early attachment relationships. Finally, Main (personal communication) has suggested that some number of “earned secures” may actually have had a positive initial foundation, with a secure attachment relationship in infancy, but adverse experience later in development. We present evidence in the next section supporting the role of early foundations as supports for positive change in adolescence.
THE DYNAMIC ROLE OF EARLY EXPERIENCE Attachment theory is a structural developmental theory, wherein subsequent development is conceived as building upon as well as transforming what preceded (Sroufe, 1996). In this systemic, dynamic position, psychopathology is viewed as a complex, organic creation, not the simple sum of positive and negative experiences. Early experience, therefore, has special significance because it frames the child’s subsequent transactions
Annual progress in child psychiatry and child development 2000–2001
44
with the environment. “The child not only interprets experience, the child creates experience. As Alfred Adler has suggested, the child is both the artist and the painting” (Sroufe, 1978, p. 57). From Freud, Bowlby drew the idea of the primacy of the earliest attachment relationships as the first experiences of emotional closeness. These vital relationships represent prototypes for close relationships throughout life, especially for intimate love relationships and parenting. Bowlby centered his own ideas on his concept of “internal working models.” Children inevitably extract from their experience expectations regarding likely behavior of others and themselves in relationships. Humans cannot keep themselves from doing this. “The varied expectations of the accessibility and responsiveness of attachment figures that different individuals develop during the years of immaturity are tolerably accurate reflections of the experiences those individuals have actually had” (Bowlby, 1973, p. 202). Although simple and straightforward, this is a profound idea. It means that children approach new situations with certain preconceptions, behavioral biases, and interpretive tendencies. Thus, context, even new circumstances and new arenas are not independent of the child’s history. As we have said elsewhere, If because of early experience the preschooler isolates himself from the peer group, he removes himself further from positive social experiences…. If selfesteem and trust are established early, children may be more resilient in the face of environmental stress. They may show poor adaptation during an overwhelming crisis, but when the crisis has passed and the environment is again positive, they may respond more quickly. Even when floundering, some children may not lose their sense that they can affect the environment…. (Sroufe, 1978, p. 45) Substantial research confirms the idea that children with varying attachment histories construe the environment differently. Such differences are revealed, for example, in their completions of stories with separation themes (Bretherton, Ridgeway, & Cassidy, 1990), their social pretend play (Rosenberg, 1984), their reactions to cartoons depicting potential social conflict (Suess, Grossmann, & Sroufe, 1992), their reactions to family photographs (Main, 1993), their family drawings (Fury, Carlson, & Sroufe, 1997; Main, 1993), and their memories for affective-cognitive stimuli (Belsky, Spritz, & Crnic, 1996; Rieder & Cicchetti, 1989). These studies show that those with secure histories are less likely to attribute hostile intent in ambiguous social situations or to reject stimuli portraying their parents, and more likely to bring fantasied conflicts to successful resolution and to see themselves as connected to others, especially family members. Although the emphasis here has been upon cognitive frameworks, it is fully compatible with recent writing on the formative impact of early experience for brain system development (e.g., Cicchetti & Tucker, 1994; Schore, 1994) and on early relationships as entraining basic patterns of emotional regulation (Sroufe, 1996, 1997). All of these levels of analysis are mutually supporting. It is also the case that expectations and biases of the child often lead to and therefore are borne out by environmental reactions in a self-perpetuating manner. For example,
Implications of attachment theory for developmental psychopathogy 45 pushing others away (even if it is at some level to avoid disappointment, see the following) frequently leads in fact to the rejection that was expected. Children with histories of avoidant attachment not only expect rejection from others, they are, indeed, often rejected by them (Sroufe, 1983; Suess, Grossmann, & Sroufe, 1992). In general, individual patterns of adaptation elicit reactions from the environment that consolidate and elaborate them. The positive feedback cycles generated by maladaptation are, in fact, a key part of its definition. From the perspective of attachment theory, then, several mechanisms underlie continuity in adaptation (Sroufe, 1988). Continuity in both environmental influences (e.g., quality of care) and individual characteristics supports stability in individual functioning over time. Prior history also supports continuity in adaptation, because continuity is viewed as a transactional process involving an active self-regulating organism and the environment (Sroufe & Egeland, 1991). From this perspective, the child actively participates in constructing his or her experience. Individuals behave in ways that elicit responses from the environment that support prior adaptation, and individuals make choices that selectively engage aspects of the environment supporting a par-ticular adaptive style. Individuals also interpret new and ambiguous situations in ways that are consistent with earlier experience. Finally, as implied in many of the preceding points, early experiences may have special significance because of their nature. Because they are preverbal, they are not accessible to verbal recall and may be less readily modified by subsequent experience. This is the essence of Freud’s idea of the “dynamic unconscious” (Loevinger, 1976). A basic sense of emotional connectedness, confidence regarding the availability of others, and feelings of self-worth may be the legacy of infancy. Although such a basic orientation may, of course, be altered by later experience, it might still be reactivated in certain circumstances or in certain domains of functioning. Specific hypotheses that may be derived from these theoretical ideas are as follows: 1. Early attachment history will have ongoing importance for later socioemotional adaptation, even after taking into account current circumstances and intermediary experiences. 2. The child’s manner of engaging the environment in subsequent developmental periods will be predictable from patterns of attachment in infancy (children, in part, create their own environments). 3. Similarly, reactions of others, including those outside the family, will be predictable from infant attachment patterns. 4. Even following change, early patterns of attachment retain a potential for reactivation. There is a homeorhetic tendency (Sameroff, 1989). 5. Certain issues and certain arenas of functioning—those tapping anxiety about the availability of others or apprehension regarding emotional closeness—are especially likely to reveal the legacy of early attachment, even during periods of generally adequate functioning. These hypotheses are difficult to test and will require decades of longitudinal research. Often quite detailed observational data are required (especially for hypothesis 5), along with broad assessments of ongoing support and life stress. Still, beginnings have been
Annual progress in child psychiatry and child development 2000–2001
46
made and some clear directions have been set. The regression analyses described in the preceding already imply that early experience adds to later assessments in predicting outcome, because attachment remains significant after later variables are entered. This is confirmed when we enter variables in reverse chronological order with attachment last (for previously published data, see Sroufe, Egeland, & Kreutzer, 1990). For example, in predicting adolescent pathology (see the preceding), quality of care at age 13 is significant, accounting for 4% of the variance. Nonetheless, when infant attachment history (avoidant and disorganized) is subsequently entered, it also is significant, accounting for an additional 14% of the variance. Early attachment also adds when the total prior equation includes behavior problems in grades one through six and quality of care at age 13 years (see Carlson, 1998). Similar results also are obtained when the outcome variable taps adolescent competence with peers. Our detailed studies of a subsample observed in nursery school, summer day camp in middle childhood, and weekend retreats in adolescence further clarify the ongoing role of early experience (e.g., Elicker, Englund, & Sroufe, 1992; Sroufe, 1983; Sroufe, Carlson, & Shulman, 1993; Sroufe, Egeland, & Carlson, in press). Two variables that have been clearly and consistently related to anxious attachment history are isolation and dependency (hovering near and/or being involved with teachers or counselors at the expense of peer interaction). Self-isolation appears to be more frequent in those with histories of avoidant attachment; orientation toward adults is more frequent in those with resistant attachment histories. Reactions of peers, teachers, and counselors to children with different attachment histories also have proven predictable. This may be due to the obvious dependency and lack of resourcefulness of those with resistant histories and the alienation and, at times, hostile aggressive behavior of those with avoidant histories. The strongest data here come from our preschool study where children were filmed daily over the course of the term. For example, from the films a random sample of at least 50 interactions for each child with each of the two head teachers was edited onto a tape. Coders blind to attachment history and all other data made ratings of dimensions of teacher treatment (Sroufe & Fleeson, 1988). For those children having secure histories, ratings were consistently higher on “expectations for compliance” and lower on “control” than for other groups. The resistant group was distinguished by high ratings on “control,” “tolerance for rule infraction,” and “nurturance,” and low ratings on “expectations for compliance.” The avoidant group elicited high ratings on “control” and low “warmth,” “expectations,” and “nurturance” ratings, and they were the only group that elicited anger. Such reactions may be seen as perpetuating the immaturity of those with resistant histories and the expectations for rejection of those with avoidant histories. The same basic picture resulted from observations in later childhood as well. In addition, peer sociometrics, and friendship selections, which were concordant for attachment history, confirmed that peer experiences also were predictable from early attachment differences (e.g., Elicker et al., 1992). One way we showed that early experience remains available even while latent is to examine groups of children who, although functioning similarly for a period are distinguished by early history. Two such groups were established during the preschool years. For both, adaptation was consistently negative across three assessments from 42 to
Implications of attachment theory for developmental psychopathogy 47 54 months, yet, members of one group had earlier histories of secure attachment. Followup study in elementary school revealed a rebound to significantly better functioning (fewer behavior problems) for those with secure histories (Sroufe et al., 1990). Thus, the resiliency of these children was predictable; their rebound toward positive adaptation was latent in the secure base that preceded the problems of the preschool period. We have shown similarly that change in behavior problems from the late elementary years to adolescence is predictable from early history. To parallel and expand on the preschool study, we created groups of children who were functioning comparably in terms of behavior problems during elementary school, but who had differed in their attachment histories. A total of four groups was created, including children with stable secure histories (at 12 and 18 months) who were functioning well or poorly in middle childhood and children with stable insecure histories who were functioning well or poorly in middle childhood (based on the CBCL in Grades 1–3).1 Behavior problem status in middle childhood was paralleled by peer competence and emotional health rankings made by teachers. For the analyses of subsequent adolescent adaptation, the critical comparisons were between the groups of children who looked the same in middle childhood, but differed in their early attachment status. These comparisons revealed that the groups differed in later psychopathology and competence in ways predictable from their early histories. Among the groups of children who were doing well in middle childhood, the ones with a secure attachment history had lower scores on present, t(44)=2.66, p<.01, and past, t(42)=2.70, p<.01, pathology indices derived from the K-SADS at age 17.5 years. They also scored significantly higher on a global rating of competence at age 19 years based on their functioning across work, school, and relationship domains, t(42)=2.91, p<.006. Similarly, for those children who were functioning poorly in middle childhood, the ones with a history of secure attachment scored lower on total, t(28)=−2.99, p=.006, and past, t(29) =2.54, p=.02, pathology on the KSADS. Thus, as suggested, individuals showing positive change in adolescence, following a period of maladaptation, drew upon a more positive foundation in the infancy period. Looking across all four groups, level of functioning in adolescence appeared to depend on both early and later experience. The children who had secure histories and who were also functioning well in middle childhood were consistently significantly higher than all other groups in their competence ratings and lower in their pathology ratings. Conversely, the children with insecure histories who had behavior problems in middle childhood were 1. Children were considered to be functioning well in middle childhood if they had total scores in the normative range (T≤ 55) on the Teacher Report Form of the Child Behavior Checklist (Achenbach & Edelbrock, 1986) for at least two of three assessments from first through third grade and also in sixth grade. Children were considered to be functioning poorly in middle childhood if they had total scores in the upper range (T ≤ 62) for at least two of four assessments in first through third and sixth grade; sixth-grade scores also had to be at least above the group mean. These criteria were established to insure consistency in behavior or adaptation across all of middle childhood.
Annual progress in child psychiatry and child development 2000–2001
48
significantlylower than all other groups in adolescent competence and higher in pathology. Notably, the two groups of children with mixed histories (secure attachment and later behavior problems or insecure attachment and later positive functioning) were comparable to each other on both adolescent competence and behavior problems. A positive early foundation appeared to be a protective factor, allowing the children to rebound somewhat from a difficult middle childhood. This finding suggests some special role for early experience, as its effects appeared after a long passage of time and seemed to be as potent as later experience in predicting adolescent outcomes. At the same time, children with insecure histories were amenable to positive change, as those who were doing well in middle childhood were also functioning fairly well in adolescence. Finally, we find that those with secure histories show less negative reaction to periods of high family stress both in middle childhood (Pianta, Egeland, & Sroufe, 1990) and adolescence (Sroufe, Levy, & Carlson, 1998). Resilience and vulnerability are best reviewed in process terms, with foundations beginning to be laid down in the earliest years (e.g., Egeland, Carlson, & Sroufe, 1993). Our data, and the literature in general, provide only hints regarding the hypothesis of special vulnerabilities or particular domains of impact that are the legacy of early experience. In recent studies of adolescent group functioning (including variables such as taking a leadership role, the capacity for interpersonal vulnerability, and self-reported intimacy in friendships; e.g., Englund et al., 1997; Ostoja, McCrone, Lehn, Reed, & Sroufe, 1995), we frankly were surprised at the degree of predictability of results from infant attachment assessments. At times these surpassed predictions to preschool and middle childhood. We propose that this may be due to the special capacities for emotional investment of self that are called for during adolescence that tap into early attachment experience in a way that earlier measures of competence do not. We are only now beginning to assess romantic relationships and parenting in the next generation, which will be crucial tests of attachment theory. Studies using the AAI (George, Kaplan, & Main, 1984), which has been linked to infant attachment history in some samples but not our own, are suggestive. Autonomy (security) on the AAI has been linked to the capacity to use a romantic partner as a secure base (Crowell, 1997) and to secure attachment in infants (van Ijzendoorn, 1995). More directly pertinent to psychopathology would be differential vulnerability to major loss experiences. In the third volume of the trilogy, Bowlby (1980) argued that early inadequate attachment history would leave individuals vulnerable to depression in the face of such losses (see also Brown, Harris, & Bifulco, 1986). Our sample has proven too small (losses of primary caregivers too few) to test this hypothesis as yet. Frequency of unresolved losses for those who are autonomous (secure) and nonautonomous on the AAI might provide a partial test.
THE NATURE OF EARLY RELATIONSHIP DISTURBANCE Bowlby’s work emphasizing the quality of early adaptation and continuity in experience provides a framework for conceptualizing early relationship disturbances and their links to psychopathology (Sroufe, 1986). From this perspective, disturbed early attachment is
Implications of attachment theory for developmental psychopathogy 49 most profitably viewed as a marker of a beginning pathological process that probabilistically leads to later pathology (Sameroff & Emde, 1989). Thus, although in the context of extreme deprivation or maltreatment, disturbance within the child may be manifested at an early age (infancy or toddlerhood), in most cases early disturbance lies within the dyadic relationship and only gradually takes the form of enduring disturbance within the child. In our view, many disturbances of childhood may be viewed as having relational origins wherein patterns of dyadic emotion regulation in time are carried forward by the child and manifested in individual styles of coping with challenge. Within this framework, anxious attachment patterns are viewed as dyadic regulatory patterns that maximize to the extent possible opportunities for the infant to maintain closeness (secure base behavior) in the context of unavailable or intermittently available and unresponsive caregiving. Avoidant infants maintain proximity to the caregiver (in case of extreme threat) by minimizing signals of distress and negativity that may alienate a rejecting caregiver (Main, 1981). For infants classified as resistant, heightened distress signals serve to maintain the attention of intermittently responsive caregivers. Even behavior of infants classified as disorganized (stereotypes, simultaneous approach/avoidance) enables infants to maintain proximity in the context of frightening caregiver behavior and internal conflict (Main & Hesse, 1990). Distortions of early dyadic regulatory processes serve as prototypes for later dysregulation, markers of a process that leaves individuals vulnerable to normative stresses and the development of pathology. Early maladaptive relationship patterns are internalized and carried forward as characteristic modes of affective regulation and associated expectations, attitudes, and beliefs. Emerging developmental capacities (cognitive, linguistic, behavioral) become organized with respect to patterns of restricted regulation or emotional dysfunction (Carlson & Sroufe, 1995). With development, individuals construct more complex and elaborated types of defensive coordination. Variations in early regulatory patterns provide the basis for differences in later strategies for coping with normative stresses, eliciting support from others and making use of internal signals (Carlson & Sroufe, 1995). For individuals for whom the caregiver has been a source of effective emotional regulation and comfort, relationships are valued. From the earliest relationship, the individual has experienced the rudiments of reciprocity, including how to receive care and respond empathically to others (Sroufe & Fleeson, 1986). The child’s expectations about self and others and consequent behavior elicit feedback that supports a particular adaptive style. Under conditions of extreme stress, the child is likely to seek comfort and support with the expectation that others will be available and will to provide aid. In contrast, individuals with histories of insecure attachment may be more likely to form relationships that are not supportive and are easily disrupted (Carlson & Sroufe, 1995). For the avoidant child, early experiences support a view of the self as isolated, unable to achieve emotional closeness, and unworthy of care. Social relationships may be viewed as alien and treated with hostility. With elevated stress, the child may fail to seek comfort from others, perpetuating a view of relationships as alien or hostile. For the child with a history of unpredictable or inconsistent caregiving experiences, an anxious style, in which negative emotions disrupt rather than restore relationships, inhibits the development of stable close relationships. Individual continuity in such patterns results,
Annual progress in child psychiatry and child development 2000–2001
50
in part, because such nonconscious, underlying processes are no longer a part of conscious social interchange and, as a result, no longer subject to environmental feedback and revision. For children with avoidant or resistant histories, emotions that would have facilitated affective communication and exchange are defensively modified or cut off (Carlson & Sroufe, 1995; Kobak, Ruckdeschel, & Hazan, 1994). As a result, when experiencing distress the child may fail to signal directly a need for support, become embroiled in negative emotion, and be unable to draw from potentially supportive social relationships. Moreover, for individuals with insecure attachment histories, working defensive strategies of avoidance or resistance may themselves be vulnerable to breaking down under stress (Kobak & Shaver, 1987). This is evident in low ratings of ego resiliency, inability to cope with frustration, and pervasive presence of negative affect of insecure children (Erickson et al., 1985; Sroufe, 1983). In the context of stressful life events, these children may be faced with emotional conflicts that their defensive strategy was organized to control but now are overwhelming (Carlson & Sroufe, 1995). Symptomatic forms of attachment behavior and other problems may result. Hypochondriacal reactions may serve needs for closeness or care, or suicidal behavior may represent a desire to reunite with a lost attachment figure (Bowlby, 1980). Other potential links between anxious attachment and adult disorder are described by Dozier and her colleagues (Dozier, Stovall, & Albus, in press). For individuals with histories of extremely harsh or particularly chaotic caregiving contexts (disorganized attachment relationships), the process of regulation, the consolidation or integration of self across behavioral states and acquisition of control over modulation of states, may be disrupted (Carlson & Sroufe, 1995). (Empirically, we have found that disorganized attachment in infancy, followed by subsequent trauma, is the most potent combination for predicting dissociation in early adulthood; Ogawa, Sroufe, Weinfield, Carlson, & Egeland, 1997.) Recurring trauma in the context of inadequate or overly restrictive caregiving increases the level of arousal and need to separate or compartmentalize overwhelming affects and memories. Dissociative mechanisms that serve survival functions by providing for the automatization of behaviors, escape from reality constraints, containment of overwhelming memories or feelings, and in extreme form the assignment of painful feelings to alternate personalities are strengthened (Ludwig, 1983; Putnam, 1989). Variations in outward manifestations result from differences in individual psychological capacities, caregiving experiences, and social and cultural factors (Ludwig, 1983). In summary, early disturbances in attachment relations, although themselves generally not best construed as psychopathology, often do lay the foundation for disturbances in developmental processes which can, indeed, lead to psychopathology. Understanding the processes wherein what begins as a relationship disturbance can in time lead to individual disorder is one of the central tasks for the field of developmental psychopathology (Sameroff & Emde, 1989).
CONCLUSION
Implications of attachment theory for developmental psychopathogy 51 Within attachment theory, psychopathology is viewed as a developmental construction, resulting from an ongoing transactive process as the evolving person successively interacts with the environment. Individual transforms environment but also is transformed by it. In this perspective early attachment variations generally are not viewed as pathology or even as directly causing pathology. Rather, varying patterns of attachment represent “initiating conditions” in systems terms. In this regard, they do play a dynamic role in pathological development because of the way in which environmental engagement is framed by established tendencies and expectations. Moreover, patterns of infantcaregiver attachment and other aspects of early experience may have a special role in the developmental process via their impact on basic neurophysiological and affective regulation. Still, anxious attachment by no means leads inevitably to psychopathology. Change remains possible at numerous points in development, although both theory and data suggest that such change is more readily accomplished early in the process or at least when there is a foundation of early support.
ACKNOWLEDGMENT Preparation of this work and the research described herein were supported by a grant from the National Institute of Mental Health (MH 40864).
REFERENCES Achenbach, T., & Edelbrock, C. (1986). Manual for The Teachers’ Report Form and Teacher Version of the Child Behavior Profile. Burlington, VT: Department of Psychiatry, University of Vermont. Belsky, J., & Isabella, R.A. (1988). Maternal, infant, and social-contextual determinants of attachment security. In J.Belsky & T.Nezworski (Eds.), Clinical implications of attachment (pp. 41–94). Hillsdale, NJ: Erlbaum. Belsky, J., Spritz, B., & Crnic, K. (1996). Infant attachment security and affectivecognitive information processing at age 3. Psychological Science, 7, 111–114. Bowlby, J. (1944). Forty-four juvenile thieves: Their characters and home life. International Journal of Psychoanalysis, 25, 19–52, 107–127. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York: Basic Books. Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss. New York: Basic Books. Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment (2nd ed.). New York: Basic Books. Bretherton, I., Ridgeway, D., & Cassidy, J. (1990). Assessing internal working models of the attachment relationship: An attachment story completion task for 3-yearolds. In M.T.Greenberg, D.Cicchetti, & E.M.Cummings (Eds.), Attachment in the preschool years: Theory, research, and intervention (pp. 273–308). Chicago: University of Chicago Press.
Annual progress in child psychiatry and child development 2000–2001
52
Bronfenbrenner, U. (1986). Ecology of the family as a context for human development: Research perspective. Developmental Psychology, 22, 723–742. Brown, G., Harris, T., & Bifulco, A. (1986). Long-term effects of early loss of parent. In M.Rutter, C.Izard, & P.Read (Eds.), Depression in young people (pp. 251–296). New York: Guilford. Carlson, E.A. (1998). A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69, 1107–1128. Carlson, E.A., & Sroufe, L.A. (1995). The contribution of attachment theory to developmental psychopathology. In D.Cicchetti & D.Cohen (Eds.), Developmental processes and psychopathology: Vol. 1. Theoretical perspectives and methodological approaches (pp. 581–617). New York: Cambridge University Press. Cicchetti, D., & Tucker, D. (1994). Development and self regulatory structures of the mind. Development and Psychopathology, 6(4), 533–549. Crowell J. (1997). Current relationship representations: Relations to secure base behavior in marital interactions. Paper presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Dozier, M., Stovall, K.C., & Albus, K. (in press). Attachment and psychopathology in adulthood. In J.Cassidy & P.R.Shaver (Eds.), Handbook of attachment: Theory, research, and clinical implications. New York: Guilford. Egeland, B., Carlson, E.A., & Sroufe, L.A. (1993). Resilience as process. Development and Psychopathology, 5, 517–528. Elicker, J., Englund, M., & Sroufe, L.A. (1992). Predicting peer competence and peer relationships in childhood from early parent-child relationships. In R.Parke & G.Ladd (Eds.), Family-peer relationships: Modes of linkage (pp. 77–106). Hillsdale, NJ: Erlbaum. Englund, M., Levy, A., & Hyson, D. (1997). Development of adolescent social competence: A prospective longitudinal study of family and peer contributions. Poster session presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Erickson, M., Egeland, B., & Sroufe, L.A. (1985). The relationship between quality of attachment and behavior problems in preschool in a high risk sample. In I. Bretherton & E.Waters (Eds.), Growing points in attachment theory and research. Monographs of the Society for Research in Child Development, 50(1–2), Serial No. 209, 147–186. Fox, N. (1997). Attachment in infants and adults: A link between the two? Paper presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Fury, G., Carlson, E., & Sroufe, L.A. (1997). Children’s representations of attachment relationships in family drawings. Child Development, 68, 1154–1164. George, C., Kaplan, N., & Main, M. (1984). Attachment interview for adults. Unpublished manuscript, University of California, Berkeley. Kobak, R.R., Ruckdeschel, K., & Hazan, C. (1994). From symptom to signal: An attachment view of emotion in marital therapy. In S.M.Johson & L.S.Greenberg (Eds.), The heart of the matter (pp. 46–71). New York: Brunner/Mazel. Kobak, R.R., & Shaver, P. (1987, June). Strategies for maintaining felt security: A theoretical analysis of continuity and change in styles of social adaptation. Paper
Implications of attachment theory for developmental psychopathogy 53 presented at the Conference in honor of John Bowlby’s 80th Birthday, London, England. Lamb, M. (1984). Fathers, mothers and childcare in the 1980s: Family influences on child development. In K.Borman, D.Quarm, & S.Gideouse (Eds.), Woman in the workplace (pp. 61–88). Norwood, NJ: Ablex. Lewis, M. (1997) Altering fate. New York: Guilford. Loevinger, J. (1976). Ego development. San Francisco: Jossey-Bass. Ludwig, A.M. (1983). The psychobiological functions of dissociation. American Journal of Clinical Hypnosis, 26, 93–99. Main, M. (1981). Avoidance in the service of attachment: A working paper. In K. Immelmann, G.Barlow, L.Petrinovich, & M.Main (Eds.), Behavioral development: The bielefeld interdisciplinary project (pp. 651–693). New York: Cambridge University Press. Main, M. (1993). Discourse, prediction, and recent studies in attachment: Implications for psychoanalysis. Journal of the American Psychoanalytic Association, 41, 245–260. Main, M., & Goldwyn, R. (in press). Adult attachment scoring and classification system. In M.Main (Ed.), Behavior and the development of representational models of attachment: Five methods of assessment. New York: Cambridge University Press. Main, M., & Hesse, E. (1990). Parents’ unresolved traumatic experiences are related to infant disorganized attachment status: Is frightened and/or frightening parental behavior the linking mechanism? In M.T.Greenberg, D.Cicchetti, & E.M. Cummings (Eds.), Attachment in the preschool years (pp. 161–182). Chicago: University of Chicago Press. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I.Bretherton & E.Waters (Eds.), Growing points of attachment theory and research. Monographs of the Society for Research in Child Development, 50(1–2), Serial No. 209, 66–104. Ogawa, J., Sroufe, L.A., Weinfield, N.S., Carlson, E., & Egeland, B. (1997). Development and the fragmented self: A longitudinal sudy of dissociative symptomatology in a non-clinical sample. Development and Psychopathology, 9, 855– 1164. Ostoja, E., McCrone, E., Lehn, L., Reed, T., & Sroufe, L.A. (1995, March). Representations of close relationships in adolescence: Longitudinal antecedents from infancy through childhood. Poster presented at the biennial meeting of the Society for Research in Child Development, Indianapolis, IN. Pearson, J.L., Cohn, D.A., Cowan, P.A., & Cowan, C.P. (1994). Earned- and continuoussecurity in adult attachment: Relation to depressive symptomatology and parenting style. Development and Psychopathology, 6, 359–373. Phelps, J.L., Belsky, J., & Crnic, K. (1998). Earned security, daily stress, and parenting: A comparison of five alternative models. Development and Psychopathology, 10, 21– 38. Pianta, R., Egeland, B., & Sroufe, L.A. (1990). Maternal stress in children’s development: Predictions of school outcomes and identification of protective factors. In J.E.Rolf, A.Masten, D.Cicchetti, K.Neuchterlen, & S.Weintraub (Eds.), Risk and protective factors in the development of psychopathology (pp. 215–235). New York: Cambridge University Press.
Annual progress in child psychiatry and child development 2000–2001
54
Putnam, F.W. (1989). Diagnosis and treatment of multiple personality disorder. New York: Guilford. Rieder, C., & Cicchetti, D. (1989). Organizational perspective on cognitive functioning and cognitive-affective balance in maltreated children. Developmental Psychology, 25 (3), 382–393. Rosenberg, D. (1984). The quality and content of preschool fantasy play: Correlates in concurrent social-personality function and early mother-child attachment relationships. Unpublished doctoral dissertation, University of Minnesota, Minneapolis. Sameroff, A. (1989). General systems and the regulation of development. In M.Gunnar & E.Thelen (Eds.), Minnesota Symposia in Child Psychology (Vol. 22, pp. 219–235). Hillsdale, NJ: Erlbaum. Sameroff, A., & Chandler, M. (1975). Reproductive risk and the continuum of caretaking casualty. In F.D.Horowitz (Ed.), Child development research (Vol. 4). Chicago: University of Chicago Press. Sameroff, A., & Emde, R. (1989). Relationship disturbances in early childhood. New York: Basic Books. Schore, A. (1994). Affect regulation and the origin of the self: The neurobiology of emotional development. Hillsdale, NJ: Erlbaum. Sroufe, L.A. (1978, October). Attachment and the roots of competence. Human nature, 1, 50–57. Sroufe, L.A. (1983). Infant-caregiver attachment and patterns of adaptation in pre-school: The roots of maladaptation and competence. In M.Perlmutter (Ed.), Minnesota Symposia in Child Psychology (Vol. 16, pp. 41–83). Hillsdale, NJ: Erlbaum. Sroufe, L.A. (1986). Bowlby’s contribution to psychoanalytic theory and developmental psychology. Journal of Child Psychology and Psychiatry, 27, 841–849. Sroufe, L.A. (1988). The role of infant-caregiver attachment in development. In J. Belsky & T.Nezworski (Eds.), Clinical implications of attachment (pp. 18–38). Hillsdale, NJ: Erlbaum. Sroufe, L.A. (1996). Emotional development: The organization of emotional life in the early years. New York: Cambridge University Press. Sroufe, L.A. (1997). Psychopathology as outcome of development. Development and Psychopathology, 9, 251–268. Sroufe, L.A., Carlson, E., & Shulman, S. (1993). The development of individuals in relationships: From infancy through adolescence. In D.C.Funder, R.Parke, C. Tomlinson-Keesey, & K.Widaman (Eds.), Studying lives through time: Approaches to personality and development (pp. 315–342). Washington, DC: American Psychological Association. Sroufe, L.A., & Egeland, B. (1991). Illustrations of person and environment interaction from a longitudinal study. In T.Wachs & R.Plomin (Eds.), Conceptualization and measurement of organism-environment interaction (pp. 68–84). Washington, DC: American Psychological Association. Sroufe, L.A., Egeland, B., & Carlson, E.A. (in press). One social world: The integrated development of parent-child and peer relationships. In W.A.Collins & B. Laursen (Eds.), Relationships as developmental context: The 29th Minnesota symposium on child psychology. Hillsdale, NJ: Erlbaum.
Implications of attachment theory for developmental psychopathogy 55 Sroufe, L.A., Egeland, B., & Kreutzer, T. (1990). The fate of early experience following developmental change: Longitudinal approaches to individual adaptation in childhood. Child Development, 61, 1363–1373. Sroufe, L.A., & Fleeson, J. (1986). Attachment and the construction of relationships. In W.Hartup & Z.Rubin (Eds.), Relationships and development. Hillsdale, NJ: Erlbaum. Sroufe, L.A., & Fleeson, J. (1988). The coherence of family relationships. In R.A. Hinde & J.Stevenson-Hinde (Eds.), Relationships within families: Mutual influences. Oxford: Oxford University Press, 27–47. Sroufe, L.A., Levy, A.K., & Carlson, E. (1998, March). Resilience in adolescence: Changing the odds. Paper presented at the biennial meeting of the Society for Research on Adolescence, San Diego, CA. Sroufe, L.A., & Waters, E. (1977). Attachment as an organizational construct. Child Development, 48, 1184–1199. Suess, G.J., Grossmann, K.E., & Sroufe, L.A. (1992). Effects of infant attachment to mother and father on quality of adaptation in preschool: From dyadic to individual organization of self. International Journal of Behavioral Development, 15(1), 43–66. van Ijzendoorn, M.H. (1995). Adult attachment representations, parental responsiveness, and infant attachment: A meta-analysis on the predictive validity of the adult attachment interview. Psychological Bulletin, 117(3), 387–403. Vaughn, B., Waters, E., Egeland, B., & Sroufe, L.A. (1979). Individual differences in infant-mother attachment at 12 and 18 months: Stability and change in families under stress. Child Development, 50(4), 971–975. Weinfield, N.S., Sroufe, L.A., Egeland, B., & Carlson, E. (in press). The nature of individual differences in infant-caregiver attachment. In P.Shaver & J.Cassidy (Eds.), Handbook of attachment: Theory, research, and clinical implications. New York: Guilford.
PART I: DEVELOPMENTAL ISSUES
4 Attachment Security in Infancy and Early Adulthood: A Twenty-Year Longitudinal Study Everett Waters, Susan Merrick, Dominique Treboux, Judith Crowell, and Leah Albersheim
Sixty white middle-class infants were seen in theAinsworth Strange Situation at 12 months of age; 50 of these participants (21 males, 29 females) were recontacted 20 years later and interviewed by using the Berkeley Adult Attachment Interview (AAI). The interviewers were blind to the participants’ Strange Situation classifications. Overall, 72% of the infants received the same secure versus insecure attachment classification in early adulthood, κ=0.44, p<0.001. As predicted by attachment theory, negative life events—defined as (a) loss of a parent, (b) parental divorce, (c) life-threatening illness of parent or child (e.g., diabetes, cancer, heart attack), (d) parental psychiatric disorder, and (e) physical or sexual abuse by a family member—were an important factor in change. Fortyfour percent (8 of 18) of the infants whose mothers reported negative life events changed attachment classifications from infancy to early adulthood. Only 22% (7 of 32) of the infants whose mothers reported no such events changed classification, p<0.05. These results support Bowlby’s hypothesis that individual differences in attachment security can be stable across significant portions of the life span and yet remain open to revision in light of experience. The task now is to use a variety of research designs, measurement strategies, and study intervals to clarify the mechanisms underlying stability and change.
This is one of three long-term longitudinal studies assessing infant attachment. See Waters, Hamilton, and Weinfield, “The Stability of Attachment Security from Infancy to Adolescence and Early Adulthood: General Introduction,” for an overall view of study design, measures, and supporting references.
Annual progress in child psychiatry and child development 2000–2001
58
INTRODUCTION One of Bowlby’s primary goals in developing modern attachment theory was to preserve what he considered Freud’s genuine insights about close relationships and development. These included insights about (a) the complexity of social, cognitive, and emotional life in infancy, (b) underlying similarities in the nature of close relationships in infancy and adulthood, and (c) the importance of early experience. To preserve these insights, Bowlby recast Freud’s insights in terms of control systems and ethological theories. He also placed his own imprint on them by replacing cathectic bonding with evolved secure base patterns as the common thread in infant and adult relationships. He also placed greater emphasis on the openness of early relationships to change, especially in light of real-life experiences. Ainsworth’s observational studies of secure base behavior at home and in the laboratory (Ainsworth, Blehar, Waters, & Wall, 1978) initially focused on normative trends in infants’ responses to novelty, separation, and reunion. Her goal was to test the appropriateness of Bowlby’s control systems model of infant behavior toward a caregiver. Subsequently, individual differences designs proved useful for examining the determinants and developmental significance of secure base behavior (Ainsworth et al., 1978; Colin, 1996). Working within Mischel’s (1968) critique of the individual differences paradigm, Masters and Wellman (1974) examined intercorrelations and stability in several studies of infant behavior in brief laboratory separations. They concluded that, consistent with Mischel’s (1968) situationist critique of the individual differences paradigm, there was little evidence of consistency in correlations across discrete “attachment behaviors” or of stability over intervals of weeks, days, or minutes. These conclusions carried considerable weight. The present study began (Waters, 1978) as an effort to clarify issues raised by the Masters and Wellman (1974) review. Strange Situation data were collected on a middleclass sample at 12 and 18 months of age. In each episode, we counted the frequency of discrete “attachment behaviors” and rated key interactive behaviors (proximity seeking, contact maintaining, proximity and interaction avoiding, and contact resisting). In addition, we classified each infant as secure, insecure-avoidant, and insecure-resistant at each age. Reliability analysis indicated that most of the discrete behaviors examined in the Masters and Wellman (1974) review were far too rare to enable us to obtain a reliable estimate of an infant’s typical behavior from brief episodes. That is, measurement failure could explain much of the negative evidence compiled by Masters and Wellman (1974). This interpretation was strengthened by evidence that stability across episodes and across time was much higher with the broader (and thus more reliable) rating scales and classifications. These results addressed the Masters and Wellman critique in detail and, in doing so, buttressed an emerging methodological defense of individual differences research (e.g., Block, 1977; Epstein, 1978). As a result, they too carried considerable weight. Lacking attachment security measures that could be applied beyond infancy, few if any
Attachment security in infancy and early adulthood 59 researchers in the mid-1970s planned long-term follow-up assessments. This obstacle was overcome with the development and validation of the Berkeley Adult Attachment Interview (Main, Kaplan, & Cassidy, 1985; see Crowell & Treboux, 1995, for a review). As Vaughn, Egeland, Sroufe, and Waters (1979) note, Bowlby’s theory predicts that secure base use and attachment representations are significantly stable across time and yet open to change in light of significant attachment-related experience. The goal of this followup study was to examine the extent of stability and change in attachment patterns from infancy to early adulthood and to stimulate research into the mechanisms underlying these developmental trajectories.
METHOD Participants and Procedure Sixty 12 month olds recruited from newspaper birth announcements in Minneapolis and St. Paul were seen in the Ainsworth and Wittig Strange Situation in 1975 and 1976. Most also participated in a 6-month follow-up at 18 months of age (see Waters, 1978). Fifty of these participants (21 males, 29 females) were relocated 20 years later and agreed to participate in the Berkeley Adult Attachment Interview (George, Kaplan, & Main, 1985). Their ages at the time of the AAI were from 20 to 22 years. As was true for their families in the original study, their socioeconomic status spanned the lower to upper middle classes. Living arrangements were diverse: 45% lived at college, 24% at home, 24% independently, 6% in other arrangements (e.g., military). Seventy-two percent described their primary occupation as “student”; 18% had completed high school and were now employed; 4% had completed college and were now employed; 6% did not mention employment. In most instances (78%) the participants’ parents had remained married. Two participants lost a parent before age 6. Two participants had a child of their own. Infant Attachment Assessment. Each participant was seen in the Ainsworth Strange Situation at 1 year of age. They were classified as secure, insecureavoidant, or insecureresistant, as described in Ainsworth et al. (1978). The insecure disorganized classification (Main & Solomon, 1986) was not yet developed when we scored these tapes. Independent coders assigned infant attachment classifications at 12 and 18 months. Each participant was classified by two independent coders; eighteen-month data were scored without the knowledge of 12-month classifications. Raters agreed on major classifications in 45 out of 50 (90%) of the cases (see Waters, 1978). Disagreements were resolved by conference. The distribution of attachment classifications at 12 months was 29 (58%) secure, 12 (24%) insecure-avoidant, and nine (18%) insecure-resistant. Adult Attachment Assessment. Adult attachment status was assessed by using the Berkeley Adult Attachment Interview (George et al., 1985) when each participant was from 20 to 21 years of age. Administration and scoring procedures are summarized in the General Introduction and detailed in Main and Goldwyn (1994). The interviews were conducted by three of the authors. Thirtyseven interviews were conducted in a private room provided by a community library; three participants were interviewed in their parents’ homes. We interviewed ten participants by telephone, nine who had moved away
Annual progress in child psychiatry and child development 2000–2001
60
from the Minneapolis area and had no plans to visit and one who was at sea with the Navy. The interviewers were blind to participants’ infant attachment classifications. Before scoring, each interview was typed, compared with the audiotape, and if necessary corrected. Two of the authors who had completed AAI training seminars conducted by Dr. Mary Main served as coders. Inter-rater agreement was assessed by using 25 of 50 transcripts. Agreement for this sample on the three major attachment classification was 84%, κ=0.72, p<.001. The distribution of AAI classifications was 25 (50%) secure, 16 (32%) insecure dismissing, and nine (18%) insecure-preoccupied. One participant in each group was classified unresolved. Negative Life Events. One of the cornerstones of Bowlby’s theory is that attachmentrelated expectations and working models remain open to revision in light of changes in the availability and responsiveness of secure base figures. That is, attachment theory predicts both stability under ordinary circumstances and change when negative life events alter caregiver behavior. To test the hypothesis that changes in attachment classification would be related to negative life events, we obtained a score on negative life events from each participant’s AAI transcript. Negative life events were defined as (a) loss of a parent, (b) parental divorce, (c) life-threatening illness of parent or child (e.g., diabetes, cancer, heart attack), (d) parental psychiatric disorder, and (e) physical or sexual abuse by a family member. The coders who counted negative life events did so without knowledge of the participants’ Strange Situation or AAI classification and without training in the AAI scoring system. To allow time for the impact of such events to be reflected in the AAI, we limited the counts to events that had occurred before age 18. To determine whether results were specific to this method of ascertaining stressful life events, we examined events reported by checklist one year later. Forty-seven completed a checklist of life events that included all of the events identified in the AAIs. This method depends less on free-recall, the manner in which interview questions are posed, the participant’s state of mind, and the amount of material produced in the AAI. These data are relevant to the present study and to the accompanying studies that obtained life events from the AAI. Participants were divided into those reporting none and those reporting one or more of the target experiences. The one or greater criterion was set a priori on the basis that all of the target experiences would be considered major life events in current research on stress and coping; each has the potential, on its own, to change expectations about caregiver availability and responsiveness. Agreement on life events classification (none versus one or more) by AAI and checklist was 78.7%, κ=0.57, p<0.002). Twenty-two participants were classified “none” and 15 were classified “one or more” by both methods. Eight were classified “one or more” by the checklist but “none” by the AAI. Two were classified “one or more” by the AAI but “none” by the checklist.
RESULTS As hypothesized, early attachment security with mother was significantly related to AAI attachment security 20 years later (see Table 4.1). Using three classifications at each age, 32 out of 50 participants (64%) were assigned to corresponding classifications in infancy
Attachment security in infancy and early adulthood 61 and early adulthood, κ=0.40, p< 0.005; τ .17, p 0.002 (AAI dependent).1 Thirty-six out of 50 participants (72%) received the same classification using the secure-insecure dichotomy, κ=.044, p<0.001; τ=0.20, p 0.002.
TABLE 4.1 Stability of Attachment Classifications from Infancy to Adulthood
Infant Attachment Classification (Strange Situation at 12 Months) Adult Attachment Classification Secure Avoidant (A) Resistant (C) (AAI) (B) Secure (F) 20 2 3 Dismissing (D) 6 8 2 Preoccupied (E) 3 2 4 Note: S/S=Strange Situation. Stability: 64% (3 groups each age) κ=0.40, p<0.005 72% (secure versus insecure) κ=0.44, p<0.001 τ (S/S dependent)=17, p 0.002 τ (S/S dependent)=0.20, p 0.002 τ (AAI dependent)=17, p 0.002 τ (AAI dependent)=0.20, 0.002. Thirty-six percent of the participants changed classification from infancy to early adulthood. Reliability and validity problems with the attachment measures certainly account for some portion of the observed change. Nonetheless, the results also suggest that experiences beyond infancy also play a role in adult security. We examined this by counting the number of attachment-relevant negative life events mentioned in each participant’s AAI transcript and relating this to whether the participant retained or changed attachment classification across age. These results are presented in Table 4.2. When mothers had reported no stressful life events, attachment stability (three groups each age) was 72%, κ=0.465, p<0.009; τ (AAI-dependent)=0.23, p 0.006. For the secure versus insecure dichotomy, stability was 78%, κ=.525, ≤.009; τ (AAIdependent)=.28, p 0.003. Hierarchical multiple regression analyses were used to determine whether (a) secure and insecure infants were equally likely to change attachment classification, (b) mothers 1. Cohen’s κ is computed from (a) the maximum level of agreement possible (100%), (b) the proportion of concordant cases (in the diagonal cells) expected by chance (from crossmultiplying marginals), and (c) the observed proportion of agreements. κ is equal to the proportion of possible agreement over and above chance that is actually obtained. In addition to the significance test associated with k, the statistic itself can be construed as an indication of effect size. To determine whether any of the present results are specific to the statistic used, we also report, where appropriate, an alternative concordance index (Goodman & Kruskal’s τ, by means of SPSS) based on a different model of chance agreement levels. When computed with AAI-dependent, τ reflects the proportional reduction in error when the Strange Situation classification is used to predict AAI classification. Complete data from which other indices can be computed are included in tables.
Annual progress in child psychiatry and child development 2000–2001
62
of secure and insecure infants were equally likely to report stressful life events, (c) infants whose mothers reported experiencing stressful life events were more likely to change attachment classification from the initial to the follow-up assessment, and (d) secure versus insecure infants whose mothers report stressful life events were equally likely to change classification. The analyses used stressful life events (presence versus absence), infant attachment classification (secure versus insecure), and their interaction to predict whether infants’ attachment classifications (secure versus insecure) changed or remained the same over the course of the study.2
TABLE 4.2 Relations of Stressful Life Events to Change in Attachment Classifications
Stability and Change from 12 Months to 21 Years Number of Stressful Retained Security Changed Security Life Events Reported Classification on AAI Classification on AAI None Total S/S sample (n=32) 25 (78%) 17 (85%) Secure in S/S (n=20) 7 (22%) 3 (15%) Insecure in S/S (n=12) 8 (75%) 4 (25%) One or more Total S/S sample (n=18) 10 (61%) 8 (39%) Secure in S/S (n=9) 3 (33%) 6 (66%) Insecure in S/S (n=9) 7 (89%) 2 (11%) Note: S/S=Strange Situation. After first entering stressful life events, R2 change for infant classification= 0.01, F (2,47)=0.50, p<.49. Thus, there was no difference in the likelihood that secure infants (31%, 9 of 29) and insecure infants (28.6%, 6 of 21) would change classification from infancy to early adulthood. After first entering infant attachment classification, R2 change for presence or absence of stressful life events=0.09, F(2,47)=4.64, p<0.037. Thus, infants whose mothers reported one or more stressful life events were more likely to change attachment classification (44.4%, 8 of 18) than infants whose mothers reported none (21.9%, 7 of 32). Finally, after both attachment classification and stressful life events were included in the analysis, the interaction term in the analysis was also significant, R2 change=0.14, F(3,46)=8.48, p<0.006. Stressful life events were significantly related to the likelihood of a secure infant becoming insecure in early adulthood (66.6% if mother reported one or more events versus 15% if she reported none, p<0.01) in secure infants. Stressful life events were not significantly related to classification changes in insecure infants. Among insecure infants whose mothers 2. The results in Table 4.2 also suggest hypotheses about changes from insecure to secure attachment in the absence of stressful life events. These deserve to be pursued with appropriate statistical power in a larger sample or meta-analysis of data from several studies. Independent assessment of stressful life events and caregiver-child interaction at several points between the initial and follow-up attachment assessments would also be useful.
Attachment security in infancy and early adulthood 63 reported one or more such events, 22% became secure as young adults versus 33.3% if mother reported none (p< 0.59). Although attachment-related stressful life events were most often associated with changes from secure to insecure attachment, this was not always the case. One participant, whose parents responded with consistent sensitive care to the childhood onset of a lifelong illness, changed from insecure to secure. The relationship between life events and attachment patterns across time was not perfect. Eight participants reported significant attachment-related stressful life events and yet retained their infant attachment status in early adulthood. Similarly, nine participants reported no such events and yet changed attachment classification.
DISCUSSION The present data provide strong evidence for the value of the secure base concept as a conceptualization of attachment relationships in infancy and adulthood. They also support Bowlby’s expectation that individual differences can be stable across significant portions of the life span. Finally, they confirm the notion that, throughout childhood, attachment representations remain open to revision in light of real experience. The success of the secure base concept as a conceptual foundation for both the Strange Situation and the AAI is important support for the notion that early and late relationships have something in common. Moreover, the present stability data support the notion that these relationships are not merely similar in kind but somehow developmentally related. Processes that may be contributing to stability include: (a) consistency in caregiver behavior across time, (b) a tendency toward persistence in early cognitive structures, (c) the relatively moderate intensity and low frequency of attachment-related stressful events in this middle-class sample, (d) the effects of individuals on their environment, and (e) stabilizing effects of personality trait variables (Waters, Kondo-Ikemura, Posada, & Richters, 1991). This study was designed to stimulate interest and help in the design of research into the roles that such mechanisms play in the consistency of attachment stability over time. A portion of the change noted in this study is attributable to measurement error. Imperfect scoring agreement introduces approximately 10% error at each age. In addition, a similar amount of error is attributable to the fact that neither the Strange Situation nor the AAI is perfectly reliable; behavior observed in a given assessment may not be entirely representative of the person’s typical behavior (see Ainsworth et al., 1978, and Crowell & Treboux, 1995, for testretest data). Correctly estimating these psychometric factors in change is important to understanding our results. Accurately assessing both stability and change is important; minimizing either would be a mistake. As Vaughn, Egeland, Sroufe, and Waters (1979) emphasized, Bowlby’s attachment theory predicts both stability and change. The portion of change in attachment classifications that proved correlated with attachment-related stressful life events provides important support for Bowlby’s ideas about (a) the openness to change of attachment representations, and (b) the importance of real-world experiences in such change. Research on the mechanisms through which
Annual progress in child psychiatry and child development 2000–2001
64
experience leads to change in attachment representations deserves high priority in current attachment research. An important conclusion from this study is that the AAI is sensitive enough to experience to serve usefully in such work. The types of events associated with change in attachment security and the underlying mechanisms of change deserve careful analysis in shorter-term longitudinal designs. Middle-class samples offer both advantages and disadvantages. They represent a large segment of the population and are ordinarily accessible, cooperative, and interested in research. This was evident in the fact that each of the participants we recontacted agreed to participate in the AAI. The educational level of middle-class participants is also an asset because the AAI makes heavy demands on a wide range of conceptual and verbal abilities. At the same time, stability in middle-class samples may reflect more than simply the inherent stability of attachment security. Both a relatively low rate of negative attachment-relevant experiences and social support structures that buffer secure base expectations against such experiences may also contribute to the stability of secure attachment in middle-class samples, just as consistent high levels of stressful events contribute to the stability of insecure attachment in disadvantaged samples. Strong social support structures might reduce the number or impact of negative experiences and thus increase stability; they could also attenuate links between negative experiences that occurred and attachment stability. The best way to address these concerns is to examine both the stability of attachment in other populations and the mechanisms of change in close detail to understand why any participant would stay the same or change. The accompanying studies provide important information about stability and change in populations with very different patterns of caregiving and life events.
ACKNOWLEDGMENTS The initial phase of this study was supported by the Young Scholars in Social and Emotional Development Program of the Foundation for Child Development. The Foundation also provided, on very short notice, critical support for the longitudinal follow-up phase. This project would not have been possible without their support, which we gratefully acknowledge.
ADDRESSES AND AFFILIATIONS Corresponding author: Everett Waters, Department of Psychology, State University of New York at Stony Brook, Stony Brook, NY 11794–2500; e-mail:
[email protected]. Susan Merrick and Leah Albersheim are in private practice in Minneapolis, MN; Dominique Treboux and Judith Crowell are at the State University of New York at Stony Brook.
Attachment security in infancy and early adulthood 65
REFERENCES Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment. Hillsdale, NJ: Erlbaum. Block, J. (1977). Advancing the psychology of personality: Paradigm shift of improving the quality of research? In D.Magnusson & N.Endler (Eds.), Personality at the crossroads: Current issues in interactional psychology. Hillsdale, NJ: Erlbaum. Colin, V. (1996). Human attachment. Philadelphia: Temple University Press. Crowell, J., & Treboux, D. (1995). A review of adult attachment measures: Implications for theory and research. Social Development, 4, 294–327. Epstein, S. (1978). The stability of behavior: I. On predicting most of the people much of the time. Journal of Personality and Social Psychology, 37, 1097–1126. George, C., Kaplan, N., & Main, M. (1985). The adult attachment interview. Unpublished manuscript, University of California at Berkeley. Main, M., & Goldwyn, R. (1994). Adult attachment rating and classification systems (version 6.0). Unpublished manuscript, University of California at Berkeley. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I.Bretherton & E.Waters (Eds.), Monographs of the Society for Research in Child Development, 50(1–2), Serial No. 209, pp. 66–106. Main, M., & Solomon, J. (1986). Discovery of a new, insecure-disorganized/disoriented attachment pattern. In M.Yogman & B.Brazelton (Eds.), Affective development in infancy. Norwood, NJ: Ablex. Masters, J., & Wellman, H. (1974). Human infant attachment: A procedural critique. Psychological Bulletin, 81, 218–237. Mischel, W. (1968). Personality and assessment. New York: Wiley. Vaughn, B., Egeland, B., Sroufe, L., & Waters, E. (1979). Individual differences in infant-mother attachment: Stability and change in families under stress. Child Development, 50, 971–975. Waters, E. (1978). The reliability and stability of individual differences in infantmother attachment. Child Development, 49, 483–94. Waters, E., Kondo-Ikemura, K., Posada, G., & Richters, J. (1991). Learning to love: Mechanisms and milestones. In M.Gunnar & L.A. Sroufe (Eds.), Minnesota symposia on child psychology: Vol. 23. Self processes and development (pp. 217–255). Hillsdale, NJ: Erlbaum.
PART I: DEVELOPMENTAL ISSUES
5 Behavioral and Physiological Responsivity, Sleep, and Patterns of Daily Cortisol Production in Infants with and without Colic Barbara Prudhomme White, Megan R.Gunnar, Mary C.Larson, Bonny Donzella, and Ronald G.Barr
To describe the behavioral and physiological responses associated with colic, the responses of 20 2-month-old infants with and 20 without colic were studied during a physical examination. Parents kept a diary of infant behaviors (including crying and fussing) for 3 days following the visit. Using Wessel, Cobb, Jackson, Harris, & Detwiler criteria, colic was defined as fussing/crying for 3 hours or more on each of the 3 days. Behavioral data coded by “blind” observers showed that during the physical exam, colic infants cried twice as much, cried more intensely, and were more inconsolable than were control infants. Despite these behavioral differences, heart rate, vagal tone, and cortisol measures indicated no appreciable difference in physiological responsivity for the two groups. At home, parents collected saliva cortisol samples at wakeup, midmorning, midafternoon, and evening for 2 days. In a finding similar to that shown by the laboratory data, the colic and control infants did not have different levels of daily average cortisol. These laboratory and home data provide no evidence of greater responsivity in the physiological substrate of difficult temperament for colic infants and are consistent with evidence of similarity in temperament once colic is resolved. At home, compared with control infants, colic infants did display a blunted rhythm in cortisol production. By diary, they also slept about 2 hours less per day than did control infants. Nighttime sleep was still significantly different when fussing/crying was statistically controlled. These data suggest that colic might be associated with a disruption or delay in the establishment of the circadian rhythm in activity of the hypothalamicpituitary-adrenocortical axis and associated sleepwake activity.
Behavioral and physiological responsivity 67
INTRODUCTION Despite its salience for parents and clinicians, the behavioral syndrome of infant colic remains largely unexplained. The cardinal symptom of colic is excessive crying that tends to increase in the first 2 months before returning to more normal and stable levels by 4 months (Barr, 1997; Gormally & Barr, 1997; St. James-Roberts, Conroy, & Wilsher, 1995; St. James-Roberts & Halil, 1991). This crying is characterized by prolonged bouts that tend to cluster in the late afternoon and evening, are difficult to soothe, and begin and end without apparent reason (paroxysmal, unpredictable; Barr, Rotmans, Yaremko, Leduc, & Francoer, 1992; Stifter & Braungart, 1992; St. James-Roberts et al., 1995; St. James-Roberts & Halil, 1991; Wessel, Cobb, Jackson, Harris, & Detwiler, 1954). Because the crying of infants with colic is understood as a signal of distress (Barr & Geertsma, 1993), continues to increase even with optimal caregiving (Barr, 1997; Miller, Barr, & Eaton, 1993; St. James-Roberts, Conroy, & Wilsher, 1998), and is associated with facial configurations thought to be related to pain (a “pain facies”; Barr et al., 1992), it is understandably taken to indicate that something is wrong or abnormal. A number of lines of evidence have led to reconsideration of this “abnormality” assumption. First, remarkably few infants with colic have been found to have an organic etiology. Cow’s milk protein intolerance is probably a real, but relatively rare, cause of colic syndrome (Barr, 1996; Gormally & Barr, 1997; Miller & Barr, 1991; Sauls & Redfern, 1997; Treem, 1994). In a recent review, Gormally and Barr (1997) concluded that organic disease accounted for 5% or less of infants with colic, depending on the setting (primary care or referral) in which it presented. Second, most, and perhaps all, of the typical symptoms of colic syndrome are found in infants without colic, although they may be more marked, frequent, or intense in infants with colic. Similarities include the typical early crying “peak” at 2 months, the evening clustering of crying, the “pain facies,” and paroxysmal, unsoothable crying bouts (Barr, 1990a, 1990b, 1997; Barr, Chen, Hopkins, & Westra, 1996; Barr et al., 1992; St. James-Roberts et al., 1995; St. James-Roberts & Halil, 1991). These similarities suggest that the colic syndrome may reflect the upper end of the normal distribution of crying. Third, although infants with colic might be perceived differently (Forsyth & Canny, 1991), may disrupt family relationships (Raiha, Lehtonen, Korhonen, & Korvenranta, 1997; Rautava, Lehtonen, Helenius, & Silanpaa, 1995), and may seem to continue to cry more even after their actual crying has decreased (Wolke, Meyer, Ohrt, & Riegel, 1994), the outcome for the infant is actually very good (Lehtonen, Gormally, & Barr, 2000; Lehtonen, Korhonen, & Korvenranta, 1994; Stifter & Bono, 1998; Stifter & Braungart, 1992). Consequently, to better explain colic syndrome in infants without organic disease, other explanations need to be considered. A leading candidate for a “nondisease” explanation is that colic is an early clinical manifestation of a more reactive, less regulated temperament (Rothbart & Derryberry, 1981). The concept of temperament as a reflection of biologically or constitutionally based differences in reactivity and regulation has gained acceptance (Rothbart, 1989). Temperamental differences, although believed to be of a constitutional or biological
Annual progress in child psychiatry and child development 2000–2001
68
origin, present early in life, and relatively stable across time and situations, may be expressed differently at different developmental stages (Goldsmith et al., 1987; Thomas, Chess, & Birch, 1968). In the first months of life, a more reactive, less regulated temperament, when extreme, may be expressed as the colic syndrome. Indeed, infants with such temperaments are expected to express more frequent and more intense negative affect, less soothability, and less predictability in responding, characteristics that often make them difficult for caregivers to manage (Rothbart, 1989). Consequently, the hypothesis that infants with colic syndrome might be infants with a constitutionally more difficult temperament is reasonable. Indeed, in a number of studies parents have reported that infants with colic have more difficult temperaments both during (Barr et al., 1992; Lester, Boukydis, GarciaColl, Hole, & Peucker, 1992) and after (Carey, 1972, 1984; Lehtonen et al., 1994; Papousek & von Hofacker, 1998; Weissbluth, Christoffel, & Davis, 1984) colic has resolved. Nonetheless, establishing the relationship between difficult temperament and colic is methodologically problematic. Because crying and fussing are predominant behaviors for both colic and difficult temperament, they are likely to be associated “by definition” when assessed concurrently. Furthermore, retrospective parent reports of temperament after colic has resolved may be biased by the experience of living with an infant who had colic. Compelling evidence for the role of temperament in colic syndrome requires measures that are independent of amounts of crying and fussing. One approach is to include physiological measures that should reflect the constitutional or biological basis of more reactive, less regulated or difficult temperament but that are independent of parent report. Reactivity refers to the threshold, intensity, or time of onset of the response, and regulation refers to the duration or rate of recovery of the response (Goldsmith et al., 1987; Lewis, 1989; Rothbart & Derryberry, 1981; Thompson, 1994). One of the advantages of conceptualizing temperament in this way is that individual differences are expected to characterize not only behavior, but also the responsive physiological systems to which they are likely related (Boyce, Barr, & Zeltzer, 1992; Lewis, 1989; Rothbart & Derryberry, 1981). Measures of heart rate, vagal tone, and salivary cortisol have frequently been used to assess physiological reactivity and regulation in association with behavioral expressions of difficult or more reactive, less regulated temperament (Gunnar, 1990). In this study, we obtained these physiological measures in 2-month-old infants undergoing a physical examination to determine whether colic was associated with greater physiological responsivity to stimulation. Another approach to understanding colic etiology is to measure other behavioral and physiological domains that might reflect differences in regulation appropriate to normal development at this age. Sleep is one such domain. A number of researchers have now reported that infants with colic sleep less than do infants without colic (Papousek & von Hofacker, 1995; St. James-Roberts, Conroy, & Hurry, 1997; Weissbluth et al., 1984). This suggests that the crying of infants with colic might be due to sleep deprivation or dysregulation. However, because crying and sleeping are mutually exclusive behaviors, disentangling sleep deprivation from excessive crying is difficult. Over the first months of life, sleep increasingly reflects day-night or circadian patterning. This maturation in day—night or circadian patterning is characteristic of many physiological systems over the early months of life (Larson, 1996). If infants with colic sleep less, and this lesser
Behavioral and physiological responsivity 69 amount of sleep is not merely a reflection of their greater crying, then we might expect other systems that have a day-night rhythm to also show evidence of disturbance or delayed maturation as well. The hypothalamic-pituitary-adrenocortical (HPA) system, as reflected in salivary cortisol production, is one such system that shows increasing circadian rhythmicity over these early months. At birth, basal production of cortisol exhibits a two-peak pattern, separated by 12 hours, but unrelated to time of day. In adults, a peak in basal cortisol production is observed in the early morning hours, and a nadir or lowest point in cortisol production is found late in the evening around the onset of nighttime sleep. This pattern of peak in the early morning and nadir in the evening is established by 3 months of age in most infants but can be observed at least as early as 6 weeks of age (Larson, White, Cochran, Donzella, & Gunnar, 1998). The early morning peak in cortisol is dampened or blunted in infants whose longest nighttime sleep bout is less than 6 hours, as recorded by parental diaries (Larson et al., 1998). Consequently, if the regulation of daily rhythms is different in infants with colic, we might expect these differences to be reflected in patterns of daily cortisol production. Because numerous studies have shown that cortisol is not necessarily increased in association with crying (Gunnar, 1992), excessive crying would not necessarily cause an alteration in the daily rhythm in cortisol production. In this study, daily fussing/crying, sleep, and awake-alert behavior were measured by parental diary, and salivary cortisol production was assessed by having parents take saliva samples from their infants at wakeup, midmorning, midafternoon, and evening. We examined whether infants with and without colic differed in sleep and daily patterns of cortisol production. We also examined whether a more marked rhythm or slope in daily cortisol patterning was associated with better maintenance of nondistressed awake behavior during the daytime hours.
METHOD Participants and Procedure Infants were eligible for the study if they had birth weights between 2,500 and 4,500 g, gestational ages between 37 and 42 weeks, a healthy prenatal history, an uncomplicated labor and delivery, and Apgar scores of 7 or greater at 5 minutes and they were singleton births. An obstetrical complications and infant general health questionnaire was adapted from Littman and Parmelee (1978) and scored for each mother-infant pair. Infants with colic were screened to ensure that they had seen their physician and that the physician had not identified an organic reason for the colicky behavior (e.g., cow’s milk protein allergy). Further, each infant’s length, weight, and head circumference were measured at the time of the laboratory visit and were determined to be within 1.5 standard scores of the mean according to a standard physical growth chart (Ross Laboratories, 1992). Thirty potential colic infants were recruited for the study through their health care providers (n=11) or a mailing and phone recruiting for a “fussy baby” study (n=19) to parents of newborns listed on county records. Of these 30 infants, 23 met the health and age criteria listed previously and the colic criteria described subsequently, and 20 had
Annual progress in child psychiatry and child development 2000–2001
70
parents who completed all components of the testing. Another 24 control infants were recruited by random phone calls to parents of infants listed in the county birth records. Approximately one in five parents contacted agreed to participate. Parents of 20 of these control infants completed all components of the testing. Demographic information on the subject sample is summarized in Table 5.1. Mothers and fathers of infants with colic were found to be slightly older than those of infants without colic (ps<.01). None of the other measures was found to differ. When a statistical adjustment was made for the number of comparisons computed to protect against Type I errors, the difference in parental age was no longer significant.
TABLE 5.1 Characteristics of Study Sample
Characteristics
Control (n=20)
Colic (n=20)
Infant Age, days 59.8±8.9 60.9±10.4 8 (40) 6 (30) Parity (first born) Sex (male) 10 (50) 10 (50) Neonatal indices 3,450 (363) 3,450 (58) Birth wt, g 45.9 (1.8) 44.7 (2.6) Birth length, cm Littman/Parmelee Score 115 (13.6) 112 (14.0) 5,510 (761) 5,242 (941) Weight at testing g Feeding type 11 (55) 13 (65) Breast only Formula only 4 (20) 3 (15) Mixed 5 (25) 4 (20) Parent Age, years Mother 30.1 (3.6) 33.6 (3.7) Father 31.2 (3.9) 35.0 (4.7) Education, years Mother 15.9 (1.9) 16.3 (1.8) Father 15.3 (1.7) 16.0 (2.1) European American Mother 20 (100) 17 (85) Father 20 (100) 17 (85) Marital status Couples 20 (100) 20 (100) Note: Results are mean ± SD or number of subjects (percent) with designated characteristics in each group. Standard deviations or percents are in parentheses. *p<.01.
Behavioral and physiological responsivity 71 Infants of parents agreeing to participate were scheduled for a laboratory visit between 9 and 10 AM to control for the circadian rhythm in cortisol production. Upon their arrival, a research assistant explained the procedures and obtained informed consent. Electrodes were then placed for cardiac recording and the infant’s mouth was swabbed to obtain a pretest salivary cortisol sample. After approximately 5 minutes of baseline cardiac recording with the infant held in the parent’s lap, the research assistant began the physical examination. The examination procedures were organized into two sets designed to be increasingly challenging and invasive. These sets were modeled after a typical wellbaby examination on the basis of previous data showing that such procedures elicit behavioral distress and elevate salivary cortisol concentrations in infants of this age (Gunnar, Brodersen, Krueger, & Rigatuso, 1996). The first stressor set—Measurement— lasted 6 to 8 minutes (M=8 min) and consisted of the parent undressing the infant and the researcher taking ear temperature, weight, length, and head circumference followed by a 2-minute soothing protocol. Note that soothing occurred even if the infant was not distressed and was terminated after 2 minutes regardless of whether the infant was soothed. The soothing protocol (Brazelton, 1984) consisted of a progression of distal (e.g., visual contact) to proximal (e.g., holding, rocking) techniques. The second stressor set—Physical Examination—lasted 15 to 18 minutes (M=16.5 minutes) and consisted of a simulated physical with bilateral ear check, stetho-scope, oral examination, and visual examination with a flashlight performed while the baby was on the parent’s lap. In addition, the infants were placed on the examination table and given the Denver II Developmental Screening Exam. At the end of this second stressor set, the research assistant made a print of the infant’s hand on a certificate for the parent to take home. Finally, the experimenter held/consoled the infant for approximately 2 minutes. These stressor sets were separated by a 7- to 8-minute (M=7.5 minutes) rest period during which the infants were alone with their parent(s). Although the soothing protocol was a component of the procedures, the experimenter followed the soothing protocol to attempt to console the infant if the infant achieved moderate to intense crying at any point in the procedure. Even if the infant could not be consoled or consoled completely, the examination proceeded, as would be the case during a well-baby examination. At the end of the laboratory session, parents were instructed in the home data collection procedures. At home they were asked to complete at least 3 days of a daily diary (see following section). On two of the days, they were asked to sample their infant’s saliva at 4 time points during the day (see following section). They also were given a temperament questionnaire to complete (Rothbart, 1981). Measures Behavioral Diary. Infant behavior in the home was assessed by means of a modified version of the Baby’s Day Diary that has been described previously (Barr et al., 1988; Hunziker & Barr, 1986) and widely used in studies of infant behavior and colic. Previous comparisons with 24-hour audiotaped recordings have indicated moderate to strong correlations between duration of audiotaped negative vocalizations and crying duration and frequency of audiotaped negative vocalization bouts and crying/fussing frequency (Barr et al., 1988; Barr, Kramer, Pless, Boisjoly, & Leduc, 1989; St. James-Roberts,
Annual progress in child psychiatry and child development 2000–2001
72
Hurry, & Bowyer, 1993). The diary also has been shown to be sensitive to colicnormal and developmental differences (Barr et al., 1992; Hill et al., 1992; St. James-Roberts et al., 1993; St. James-Roberts & Plewis, 1996) and to interventions that affect crying and fussing behavior (Barr, Wooldridge, & Hanley, 1991; Hunziker & Barr, 1986). Total daily fussing/crying behavior was used to classify infants with and without colic according to modified Wessel’s criteria (Lehtonen et al., 2000; Wessel et al., 1954). In congruence with other researchers, an infant was considered to have colic if he or she fussed, cried, or both for 3 hours or more on each of the 3 days (Barr et al., 1992; Lehtonen et al., 1994; St. JamesRoberts et al., 1995). The range of daily fussing/crying averaged over the 3 days was 3 to 7 hours for the colic infants (M=4.88 hours) and .25 to 2.5 hours for the control infants (M=1.47 hours). In addition, the diary provided totals for sleep and for nondistressed awake behavior that were analyzed for this report. Each diary day was divided into three 8-hour blocks starting at midnight and labeled nighttime, daytime, and evening, respectively. These blocks were chosen to allow reflection of the evening peak in crying typically noted between 1,600 and 2,400 hours (Hunziker & Barr, 1986). This evening peak is poorly identified when 6-hour blocks, more typically used by researchers studying sleep, are employed (St. JamesRoberts et al., 1997). Sleeping and crying times were totaled within each block, and the arithmetic mean of 3 days was calculated for each. Awake, noncry/ fuss behavior was analyzed for the daytime hours (8 AM to 4 PM). Because colic and noncolic groups are defined by amounts of crying and because crying and nondistressed behavior are mutually exclusive, nondistressed awake behavior was included only in the analyses performed within groups (i.e., correlations among measures within groups). The diaries were each tabulated by two research assistants independently. Two discrepancies were noted and corrected. Parent Perceptions of Temperament. Parent perceptions of infant behavior were assessed by using Rothbart’s (1981) Infant Behavior Questionnaire. On the basis of evidence that Distress to Limitations reflects the core of parental perceptions of difficult temperament, we focused our analyses on this scale (Bates & Bayles, 1984). Distress to Limitations is a 20-item scale (e.g., when held or fed, how often did your baby squirm or turn her body away?) with seven response options (never to always) for each item. Laboratory Fussing/Crying. The laboratory sessions were videotaped from an adjacent room. Four research assistants who were unaware of the group status of each infant scored the videotapes. Four levels of negative vocalization were scored in 15-second intervals by using the Observer Software System (The Observer, 1993). In each interval, the predominant level of negative vocalization was scored. The levels were: (a) fussing, defined as discrete squeaks, frets, gasps/sighs with increased motor movement equivalent to Brazelton’s (1984) active awake/fussy state; (b) minimal cry, defined as longer crylike vocalizations than fussy, with more motor movement and body tension; (c) moderate cry, defined as a rolling cry, louder than a minimal cry but still consolable; and (d) intense cry, defined as a rolling cry, often with intermittent sharp inhalation, screaming, or both, more difficult to soothe, with face and body often flushed. Interrater agreement was examined by having the coders overlap on 15% of the sample, roughly divided between colic and control infants. Cohen’s κ for behavioral observations between all four coders averaged 0.71 and was greater than 0.65 for each possible pairwise comparison.
Behavioral and physiological responsivity 73 Proportion scores then were computed for each of the four periods of testing (baseline, measurement, rest, and physical examination). These scores reflected the number of 15second intervals for which each level of negative vocalization was tallied, divided by the number of intervals in that period. Thus we were able to adjust for small variations in the timing of each period between subjects. Because some levels of negative vocalization did not occur during some periods of the test procedure, two types of scores for fussing and crying were derived. First, the Total Negative Vocalizations (TNV) score was computed for each segment of the test period by summing the proportion scores for each level of negative vocalization within that period. This score usefully eliminated the problem with zero frequencies but also lost information about the intensity of crying. To examine the intensity of negative vocalizations, the proportion of each level of negative vocalization was determined for the entire test period. This proportion was then weighted from 1 (fussing) to 4 (intense crying), summed, and divided by 4 to yield a summary Intensity of Negative Vocalizations (INV) score. Laboratory Consolability. As noted, infants who achieved moderate to intense crying for 15 seconds or longer were soothed during the examination according to a standard soothing protocol (Brazelton, 1984). Each file was reviewed to determine which infants had achieved the criteria for crying that required the soothing protocol. Twenty-six cases were noted. Videotapes for these 26 infants were viewed and scored for consolability. A consolability scale was developed on the basis of previous scales by Brazelton (1984) and Riese (1995), where 0=inconsolable with maximal caregiver effort, 1=consolable with maximal caregiver intervention, 2=consolable with holding and rocking, 3=consolable with picking up, 4=consolable with touch without picking up, and 5=consolable with only visual regard and talking by the caregiver. Because infants varied in the number of times they could be scored for consolability, these scores were averaged and each infant who reached the crying criteria received an average consolability score. One of us rescored 15% of the consolability tapes for interrater reliability. Percent agreement was 100%. Salivary Cortisol. In the laboratory, three samples of saliva were obtained for cortisol determination: (a) Baseline (when the infant arrived for testing), (b) midexamination (approximately 25 minutes after arrival and at the beginning of physical examination) reflecting adrenocortical activity during measurement, and (c) postexamination (approximately 50 minutes after arrival and 20 to 30 minutes from the beginning of physical examination) reflecting the cumulative impact of the testing procedures. The collection times were as follows: baseline between 8:50 and 10:20 AM (M=9:36 AM), midexamination between 9:24 and 10:45 AM (M=10:00 AM) and postexamination between 9:50 AM and 11:30 AM (M=10:27 AM). Saliva was collected by using 3-inch cotton dental rolls to swab the infant’s mouth. The wet portion of the cotton roll was then placed in the barrel of a needleless syringe and expressed into an airtight vial for storage at −20°C. If the infant had milk within 15 to 20 minutes of the sample, the mouth was swabbed before sampling to remove any residue. Parents were instructed in the saliva collection procedures and were sent home with sufficient materials to collect eight samples, four per day. They were instructed to collect these samples on 2 of the 3 days that they kept the behavior diary. Times for collection
Annual progress in child psychiatry and child development 2000–2001
74
were 15 to 30 minutes after the infant’s morning wakeup defined as waking between 6 and 9 AM (M=7:18 AM, SD=42 minutes), midmorning between 10 and 11 AM (M=10:19 AM, SD=29 minutes), midafternoon between 3 and 5 PM (M=4:05 PM, SD=30 minutes), and evening before putting the infant down for the night (M=8:10 PM, SD=36 min). There were no significant differences by t-test between groups in the timing of any of the collections. Salivary cortisol assays were performed in batches with all samples from each baby in the same assay batch. A minimum of two control saliva vials were included in each batch to establish consistency within and between assay batches. Assays were performed in duplicate by using a modification of the Pantex Corporation cortisol RIA kit. The intraassay and interassay coefficients of variation were less than 10%. The correlation among duplicates was >0.98. Because cortisol data are typically positively skewed, the data were log 10 transformed. None of the infants was missing any of the laboratory cortisol samples. At home, however, a number of vials contained insufficient saliva (<100 µL) for assay. The home values were therefore averaged over the 2 days of collection to reduce data loss owing to insufficient sample volume. Even so, 2 control and four colic infants failed to have sufficient volume in enough of their samples to calculate levels for each of the 4 time points. To see if those without home data differed from those with home data, the behavior data (crying and fussing) from the diary were examined by using a 2 (sufficient versus insufficient samples)×2 (control versus colic) analysis of variance (ANOVA). Infants lacking complete cortisol data from the home assessment were not different from those with complete cortisol data (p=.50). As a reflection of the daily rhythm in cortisol production and on the basis of procedures by Smyth and colleagues (1997), time of day was regressed against the log-transformed cortisol values and the slope of the regression was determined for each infant separately. A higher peak morning level with a more marked and consistent decline in cortisol concentrations over the day was reflected in a more negative slope. Cardiac Measures. We acquired cardiac data by using the Vagal Tone Monitor-II (Delta-Biometrics Inc., 1994). Raw ECG data were recorded from disposable Ag/AgCl sensors placed in a triangular configuration on the chest. The signal was digitized at 1 kHz and was bandpass filtered at 5 Hz and 200 Hz, with a 3-dB cutoff. A computer algorithm detected R-spikes from the raw wave, and interbeat intervals were computed. We used MXedit software to visually display the heart period data, edit outliers, and quantify the heart period and Vagal Tone Index (Porges, 1985). The Vagal Tone Index assesses the amplitude of the respiratory sinus arrhythmia (RSA), which provides a description of parasympathetic influences on the heart (Porges, 1995). Measures of vagal tone and heart rate (heart period divided by 60,000) were calculated for each of the four periods of the laboratory test: baseline, measurement, rest, and physical examination. We established a priori decision rules for valid and reliable data. First, bradycardic events that were sufficiently large (>50% decrease from modal heart rate) were removed from the data analysis. Second, at least 30 seconds of continuous, artifact-free data were needed to calculate vagal tone for any period of laboratory testing. Five infants (3 control and 2 colic) were missing data from at least one period and thus were not included in the cardiac data analysis. We computed a 2 (missing versus not
Behavioral and physiological responsivity 75 missing cardiac data)×2 (control versus colic) ANOVA on sum of the negative vocalizations from the laboratory test period. The results for the main effect of missing cardiac data were significant, F(2, 38)=4.93, p<.05. Unexpectedly, the infants with missing cardiac data had fewer negative vocalizations in both the control (Ms 0.4 versus 1.06, SDs=.42 and .58) and colic (Ms=0.95 versus 1.66, SDs=.84 and .75) groups. Data Analysis We initially performed all of the analyses described in the next section by using gender as a factor. Neither significant main nor interaction effects of gender were noted for any of the analyses. Thus, to preserve power, we recomputed the data analyses by collapsing across gender. We analyzed behavior data by using proportion scores as raw scores and after natural log transformation. Because none of the results differed in significance, we have presented results based on raw scores. To analyze group differences, we used a t-test for questions based on individual variables, multivariate analysis of variance (MANOVA) when multiple variables were involved, and repeated measures analysis within MANOVA programs (SPSS) for measures assessed across trials. We used the Mauchly test (Norusis, 1994) to examine repeated measures data for sphericity. When sphericity was significant, we used GreenhouseGeisser Epsilon to correct for it. We performed post hoc tests by using tests for simple effects to examine effects over trials within groups. We used NeumanKeuls to examine differences among trials and Pearson correlation coefficients to examine associations among variables. To control for baseline, we established correlations with physiological response measures by using difference scores computed by subtracting the relevant baseline data.
RESULTS Responses in the Laboratory In the first set of analyses, we examined behavioral and physiological responses in the laboratory. These analyses proceeded as follows. First, we examined the data on negative vocalization and consolability to determine whether the test situation elicited distressed behavior from the infants and whether the colic infants were more distressed and less consolable than controls. Next, to determine whether our test situation elicited distressed behavior comparable to that seen at home, we examined the associations between negative vocalizations in the test situation and the home diary data on fussing/crying and parental reports of infant distress to limitations. We then analyzed group differences in physiological reactivity to the physical examination and associations between physiological and behavioral responsivity. Behavioral Responses: Crying. The proportion of time infants expressed negative vocalizations by periods of the test procedure is shown in Figure 5.1. For both groups combined, negative vocalizations increased in the stressor periods relative to the cardiac baseline period, Trial F(3, 114)=3.37, p<0.001.
Annual progress in child psychiatry and child development 2000–2001
76
Figure 5.1. Proportion of time infants expressed negative vocalizations by periods of the test procedure. Bars reflect standard errors of the mean. Note that baseline refers to cardiac baseline occurring after electrode placement, not to behavioral baseline before any manipulations. Post hoc tests indicated that negative vocalizations were more frequent during both the measurement and physical examination periods than during cardiac baseline, ps<0.05, and between the physical examination stressor compared to the measurement stressor periods, p<0.05. Colic infants exhibited more negative vocalizations than did control infants, groups F (1, 38)=8.67, p<0.001. This difference between control and colic infant did not depend on the period of the test, Interaction F(3, 114)=1.67, ns. Thus the test situation elicited negative vocalizations for both colic and control infants. Overall, colic infants fussed and cried about twice as much as did controls (see Figure 5.1). Infants with and without colic differed in intensity as well as amount of negative vocalizations. Figure 5.2 (left panel) shows group differences in the intensity of crying averaged over the entire test period (INV score). The crying of infants with colic was judged to be significantly more intense than that of infants without colic, t(38)=2.96, p<0.01. Figure 5.3 illustrates the percentage of negative vocalizations scored as fussing, minimal, moderate, and intense crying. As previously reported (Barr et al., 1988; St. James-Roberts et al., 1995), fussing was the most frequent type of negative vocalization for both colic and control infants.
Behavioral and physiological responsivity 77
Figure 5.2. Group differences in intensity of crying, consolability, and distress to limitations. Values have been converted to standard scores for the purposes of explication. Bars reflect standard errors of the mean.
Figure 5.3. Proportion of negative vocalizations scored as fuss, minimal cry, moderate cry, and intense cry.
Annual progress in child psychiatry and child development 2000–2001
78
Behavioral Responses: Consolability. Sixteen infants with colic and 10 infants without colic reached levels of negative vocalizations intense enough to be scored for consolability. As shown in Figure 5.2 (middle panel), the control infants were judged to be more consolable than the infants with colic, t(24)= 3.19, p<0.01. We examined the individual scores for consolability. Eight of the 16 colic infants received scores that indicated that they either failed to console or consoled only briefly with maximal intervention. None of the control infants received scores in this inconsolable range. Comparison of Laboratory to Home Behavior. Parents described infants with colic as exhibiting more distress to limitations than infants without colic, t(38)=7.1, p<0.01 (see Figure 5.2, right panel). In addition, all three behavioral measures in the laboratory (e.g., amount of negative vocalizations, intensity of crying, and consolability) were correlated with both parent reports of distress to limitations (rs ranged from 0.47 to 0.57, ps<0.01, with a negative sign for consolability, as would be expected) and with parent diary values for total daily crying (rs ranged from 0.44 to 0.48, ps<0.01, again with a negative sign for consolability). Thus, crying and consolability observed in the laboratory were consistent with the variations parents noted in the home. The magnitude of the associations indicated good cross-situational stability and cross-method agreement in distressed behavior. Physiological Responses in the Laboratory. Responses to the stressor experience were noted for all three of the physiological systems (Figure 5.4). Overall, heart rate increased significantly over the examination period, trials F(3, 99)=6.5, p<0.01 (see Figure 5.4A). There was some evidence of a differential increase for the infants with colic compared to control infants, interaction F(3, 99)=3.1, p<.05, that was due to infants with colic increasing their heart rate over the test period, simple effects F(3, 48)=5.7, p<.01, whereas control infants did not, simple effects F(3, 51)=1.6, ns. By t-test, however, heart rate did not differ significantly between infants with and without colic during any period of the test, ps>0.10. Inspection of Figure 5.4A suggests that the difference was due to infants with colic starting at a slightly, but not significantly, lower baseline and rising to a slightly higher level, without the levels actually differing statistically at any time point. Vagal tone also exhibited a response to the examination period, trials F(3, 99)=4.4, p<0.01 (see Figure 5.4B), increasing from cardiac baseline to measurement, p<0.05, and decreasing from measurement to physical examination, p>.05. However, no difference between infants with and without colic was noted, group F(1, 33)=0.95, ns. Cortisol also increased over the examination period, trials F(2, 76)=7.2, p< 0.01 (see Figure 5.4C), with levels higher after the physical examination than after the measurement period. Infants with and without colic did not differ in cortisol at any point in the test period, groups F(1, 38)=1.35, ns. In summary, in contrast to the group differences in behavioral responses, there was remarkably little evidence of group difference in any of the physiological responses measured during testing. We then examined the correlations between behavioral and physiological responsivity to the examination. Because there were group differences in behavioral but not physiological responses, we first computed these correlations for each group separately. However, the direction, magnitude, and pattern of significant associations were essentially the same for both groups. For neither group were significant associations found between baseline measures of heart rate, vagal tone, and cortisol and distressed
Behavioral and physiological responsivity 79
Figure 5.4. Heart rate (A), vagal tone (B), and salivary cortisol (C) shown by group and by sampling period. Bars reflect standard errors of the mean. behavior during the physical examination period, rs<0.32, ps>0.10. For the groups combined, significant correlations were found between the intensity of distress and physiological responsivity during the measurement period, δ heart rate, r(34)=.56, p<.01; ? vagal tone, r(34) =−0.41, p<0.05; ? cortisol, r(39)=0.32, p<0.05. During the physical examination period, the various measures of distress were fairly closely associated with the measures of physiological responsivity. This is illustrated in Table 5.2. In this table, the correlations are shown for the groups combined and separately by group. The similarity in magnitude and patterning of associations for infants with and
Annual progress in child psychiatry and child development 2000–2001
80
without colic suggests that despite the fact that the infants with colic have greater behavioral response to the examination stressor, the physiology-behavior organization of their responsive systems is similar to that of those infants who do not have colic.
TABLE 5.2 Correlations between Behavioral Reactivity and Physiological Reactivity during the Examination Period Computed for the Groups Combined and Separately for Control and Colic Subjects
Behavioral Reactivity Amount of Negative Intensity of Negative Consolabili Vocalizations Vocalizations Physiological Reactivity Combined Control Colic CombinedControl Colic CombinedContro ∆ Heart rate 0.80** 0.72** 0.79** 0.77* 0.73* 0.70* −0.55** −0.75* ∆ Vagal tone −0.69** −0.66**−0.71** −0.66* −0.69* −0.58* 0.34 0.30 ∆ Ig10 0.15 −0.05 0.29 0.29*** 0.10 0.32 −0.26 −0.31 Cortisol Note: Delta heart rate and vagal tone ns=18 control and 17 colic, N=35. Cortisol ns control, 20 colic, N=40. Consolability ns=10 control, 16 colic, N=26. *p<0.05; **p<0.01; ***p<0.10. Behavioral and Physiological Rhythms in the Home Sleep-Awake Behavioral Rhythms. Overall percentage of time sleeping during 8-hour periods of the day (day, evening, nighttime) are shown in Figure 5.5. Sleeping varied across the day, trials F(2, 76)=190.9, p<.001, with sleeping being recorded during a larger percentage of time at night than during the day or evening, ps<0.05. As expected, infants with colic slept less than did control infants, groups F(1, 38)=35.9, p<0.001. Control infants slept about 14 hours/day (SD=3.2 hours), whereas infants with colic slept 11.8 hours/day (SD=3.5 hours). Crying and sleeping were strongly related for infants with colic, with infants who cried more being the ones who slept less, nighttime r(18) =−0.60, daytime r(18) =−0.63, evening r(18) =−0.76, ps<0.01). This was not the case for control infants. For the control infants, knowing how much an infant cried was not a good predictor of how much they slept or vice versa, nighttime r(18) =−0.27, daytime r(18) =−0.25, evening r(18) =−0.35, ns. Thus, when not asleep, infants with colic tended to be crying, whereas this was not necessarily true for infants without colic. Data for the control infants indicate that a close inverse relationship between sleeping and crying is not a necessary relation at this age. Nonetheless, we cannot rule out the possibility that the difference in total sleep for infants with and without colic was merely due to the mutually exclusive nature of crying and sleep for the infants with colic. To determine whether there were sleep differences independent of crying differences, we compared the groups for differences in sleeping after controlling statistically for crying (MANCOVA; Figure 5.5). Infants with colic still tended to sleep less than control infants, groups F(1,
Behavioral and physiological responsivity 81 38)=3.27, p=0.08, but the difference was significant only during the nighttime, interaction F(2, 75)= 5.05, p<0.05; simple effect of groups at nighttime, F(1, 38)=4.1, p=0.05.
Figure 5.5. Proportion of time sleeping by group. Means based on raw scores in the left panel along with standard error bars. Means after adjusting for crying shown in right panel along with standard error bars. Cortisol Daily Rhythms. To determine whether infants with colic might show other alterations in circadian systems that would not be confounded with cry-ing, we examined the daily pattern in salivary cortisol production (see Figure 5.6). Analyzed by using the log 10 values, cortisol levels decreased over the daytime hours, F(3, 96)=13.0, p<0.01. Averaged over the day, infants with colic did not have higher cortisol levels than control infants, F(1, 37)=1.39, ns. Nonetheless, change over the day was dependent on colic status, interaction F(3, 96)=2.94, p=0.05. Infants with colic had a less clearly defined daily rhythm in cortisol production than did infants without colic. That is, the cortisol slope was flatter for infants with colic (M=−0.17, SD=0.36) than for controls (M=−0.50, SD=0.66), t(32)=1.68, p=0.05, where a more negative slope indicates a more clearly marked daily rhythm. To examine whether a more marked daily rhythm in cortisol production was related to infant behavior during the daytime hours, we computed correlations after reverse scoring cortisol slope so that a positive score would indicate a more marked daily rhythm. We computed the correlations within each group so that these associations would not merely
Annual progress in child psychiatry and child development 2000–2001
82
restate the differences in cortisol slope and behavior already reported. We noted no significant associations for infants with colic. However, for the infants without colic, a more marked daily rhythm in cortisol production was associated with less sleeping during the day, r(16)=−0.56, p<0.05, and more time spent in awake, nondistressed states, r(16) =0.58, p<0.05.
Figure 5.6. Salivary cortisol levels (reconverted from log 10 for purposes of explication). Bars reflect standard errors of the mean. DISCUSSION The behavioral results of this study are remarkably consistent with previous descriptions of the behavior of infants with colic syndrome as defined by modified Wessel’s criteria. By parental diary, these infants cried and fussed an average of 4.88 hours/day, and their mothers rated them as temperamentally more reactive or distressed to limitations than infants without colic. In addition, they were also behaviorally more distressed during our laboratory session, thereby providing an independent confirmation of the maternal questionnaire and diary report. Even when handled by the trained occupational therapist who served as our tester, infants with colic cried an average of twice as much as infants without colic and cried more intensely, and many proved difficult if not impossible to soothe. We determined colic status from diary measures obtained in the days following examination. Research assistants who were blind to the “colic” status of the infant and to the purpose of the study coded the behavioral measures from videotape. All of this suggests that the crying of these infants with colic was not simply due to inexpert handling by their mothers, nor did the colic appear to be in the eye of the beholder. Despite these differences in behavior and the apparent stability of infant distressed behavior across contexts (laboratory and home), the results yielded little evidence that infants with colic exhibit different patterns of physiological responsivity than do other
Behavioral and physiological responsivity 83 infants. This was the case even though heart rate and cortisol all increased markedly during the laboratory test and vagal tone decreased in the latter, more invasive, part of the procedures. These changes in physiological responding were consistent with our expectation that the examination procedures would elicit a stress response of the autonomic and adrenocortical systems. Thus, infants with colic were more behaviorally distressed than infants without colic, but they showed the same autonomic and neuroendocrine stress reactions. Of course, we cannot rule out the possibility that some other measure of physiological responsivity might have shown a difference; nonetheless, the lack of a difference on these standard indices of responsivity provides no support for the hypothesis that colic is an early manifestation of more reactive, less regulated temperament. At first glance, these findings seem inconsistent with reports that infants with colic have more difficult temperaments both during (Barr et al., 1992; Lester et al., 1992) and after (Carey, 1972, 1984; Lehtonen et al., 1994; Papousek & von Hofacker, 1998; Weissbluth et al., 1984) the period of active colic symptoms. However, the findings of the present report are actually not that inconsistent with the literature. As noted, concurrent reports of associations between colic and temperament are inherently confounded because of the overlap of symptoms (crying). Indeed, we also found differences on the temperament scale (distress to limitations) between infants with and without colic, although we did not find differences in physiological responsivity. Associations noted after colic symptoms have subsided have their own problems. Most challenging is the fact that parents have been shown to continue, inaccurately, to perceive infants with colic as crying or reacting more to stimulation after the colic symp-toms have subsided (Wolke, Meyer, Ohrt, & Riegel, 1995). This may explain why Lehtonen and colleagues (1994) found that at 1 year of age, parents were still describing infants who had exhibited colic as “more difficult.” However, they only did this on the question that asked for global impressions of the infant and not on any of the scales that requested more behaviorally grounded responses. More telling are the studies that have obtained observational measures of infant behavior before or after (or both before and after) colic has resolved. In one report, a temperament index taken at 2 weeks accounted for only 7% of the variance in crying and fussing at 6 weeks (Barr et al., 1989). In another study, there were no differences on observational measures of behavioral reactivity/distress in the newborn period or at 5 or 10 months in infants who had colic in the second month of life (Stifter & Braungart, 1992). Thus, there is actually only weak evidence that colic is an early manifestation of reactive or difficult temperament. The lack of a difference in physiological responsivity during the examination also may bear on a recent, counterintuitive hypothesis. Barr (1998) has argued that rather than serving as an indication that something is wrong or abnormal, the cry of the infant with colic may serve as an “honest signal” of health and robustness. Crying is considered to be metabolically costly. In newborns, at least, energy expenditure is increased by about 13% in the crying relative to the awake, alert state (Rao, Blass, Brignol, Marino, & Glass, 1997). In the present sample, infants with colic cried on average for 3.4 hours more per day than did infants without colic. If their greater crying involved an equivalently greater activation of cardiac, autonomic, and neuroendocrine systems, then the energy needs of these infants would be much greater than those of infants without colic. Indeed, it has
Annual progress in child psychiatry and child development 2000–2001
84
long been surprising that infants with colic gain weight equivalently to infants without colic despite all the hours they spend crying (Barr, McMullan et al., 1991; Barr et al., 1992; Lehtonen, 1994; Wessel et al., 1954). However, if our data are correct, infants with colic may be able to sustain periods of intense crying at lower levels of autonomic and neuroendocrine activity than would be expected. Certainly, this is what happened during our laboratory testing. Also, when at home, infants with colic cried for more hours per day than infants without colic but did not average higher levels of cortisol. Turning now to the home diary data, as expected, infants with colic slept less than controls. Sleep differences were noted during all time blocks: daytime, evening, and nighttime. Over the course of a day, infants with colic slept 1.5 to 2.0 hours less than controls. In the group with colic, but not in controls, infants who cried more also slept less. Because these behaviors were mutually exclusive, the sleep difference for colic infants might merely reflect their greater cry behavior. However, when we statistically controlled for crying, we still noted a statistically significant sleep difference during the nighttime (midnight to 8 AM). This is consistent with previous reports that infants with colic sleep less (Papousek & von Hofacker, 1995; St. James-Roberts et al., 1997; Weissbluth et al., 1984) and further suggests that at least some of this sleep difference is over and above that accounted for by the increased crying. We also noted differences in the daily rhythm in cortisol production between infants with and without colic. In a previous study we found that a significant peak in cortisol in the early morning was already apparent in infants as young as 6 to 7 weeks of age (Larson et al., 1998). The results of the present study were similar. Control infants, who averaged around 8 weeks of age, showed a clear daily pattern in cortisol production. Furthermore, control infants with a more marked daily rhythm or slope in cortisol production also slept less during the daytime hours and spent more of these daytime hours in awake, nonfussy states. Although these correlations might reflect a dimension of individual difference, they might also reflect the development of coordination among the different behavioral and physiological systems that support day-night rhythmicity. Although control infants exhibited a clear and marked daily rhythm in cortisol, infants with colic did not. They exhibited a blunted cortisol rhythm that was not correlated with daytime behavior. This blunted rhythm might reflect a delay in the maturation of the cortisol rhythm, or a suppressive effect of prolonged crying and decreased sleep. Without examining cortisol and crying over the period when colic resolves, we cannot tell whether the emergence of a more marked rhythm in daily cortisol production precedes or follows the resolution of colic behavior. Of course, a blunted rhythm in cortisol production may also be a characteristic of infants with colic that will not change when colic symptoms wane. A more speculative interpretation is that these differences reflect individual differences in the coordination of semiautonomous circadian rhythms in the first 3 months of life. The coordination and consolidation of day-night rhythms in behavior and physiology is one of the major developmental tasks of this age (Coons & Guilleminault, 1982; McMillen, Kok, Adamson, Deayton, & Nowak, 1991). Multiple oscillator systems regulate circadian rhythms (Anders, 1982; Moore-Ede, 1986). By 6 weeks of age most infants exhibit their longest period of sleep during the nighttime hours (Coons & Guilleminault, 1982). However, not until about 3 months of age is REM sleep clearly
Behavioral and physiological responsivity 85 consolidated into nighttime sleep and then increasingly shifted towards the later hours of nighttime sleep (Coons, 1987). Other systems increasingly exhibit circadian organization over the second and third months of life, but the emergence of circadian rhythmicity in systems regulated by one type of oscillator is relatively uncorrelated with the emergence of day-night activity in systems regulated by other oscillators. As a result, infants may experience a developmentally normative phase of relative incoordination among different circadian systems. This normative period of “heterochrony” may make it difficult for some infants to maintain nonfussy behavior during awake periods. According to this hypothesis, the typical increase in crying duration of infants between 6 and 8 weeks reflects this normative period of heterochrony among behavioral and physiological circadian systems. Colic may result when this heterochrony is greater, perhaps because of differential timing of the development of circadian activity in some systems. Colic and the normative increase in crying wanes as the various circadian systems become better coordinated and integrated over the third and fourth postnatal months. An evaluation of this hypothesis would require assessment of the many systems that exhibit emergence and coordination of periodicity with the day-night cycle. Importantly, however, although the daily patterning of cortisol differed for colic and noncolic infants, as noted, the level of cortisol production averaged over the day was comparable. This was true despite the fact that infants with colic, by definition, cried more than infants without. Thus, like the laboratory data, the home data suggest that infants with colic appear to sustain their greater crying without greater activation of systems that modulate energy metabolism or reflect the physiological substrate of more reactive, less regulated or difficult temperament.
ACKNOWLEDGMENTS The authors wish to thank the parents and infants who helped with this research. Thanks are also expressed to Renner Anderson, Judson Reaney, and the staff at Park-Nicollet Medical Center for their help in recruitment, and to Pat Larson and Mary Fowler of the Endocrine Laboratory at the University of Minnesota for their careful analysis of the salivary cortisol data. This research was supported by a grant from the National Institute of Child Health and Human Development (HD16494) and a National Institute of Mental Health Research Scientist Award (MH00946) to Megan R.Gunnar and also by the Louis Sessenwein Trust Academic Award, the Medical Research Council of Canada (MT-13633), and the Hospital for Sick Children Foundation (XG96–003) grants to Ronald G.Barr. Portions of these data were presented at the International Society for Infant Studies, Providence, RI, 1996.
ADDRESSES AND AFFILIATIONS Corresponding author: Megan R.Gunnar, Institute of Child Development, University of
Annual progress in child psychiatry and child development 2000–2001
86
Minnesota, 51 East River Road, Minneapolis, MN 55455; email:
[email protected]. Mary C.Larson and Bonny Donzella are also at the University of Minnesota; Barbara Prudhomme White is at the University of New Hampshire, Durham; and Ronald G.Barr is at the McGill UniversityMontreal Children’s Hospital Research Institute, Montreal, Canada.
REFERENCES Anders, T.G. (1982). Biological rhythms in development. Psychomatic Medicine, 44, 61– 72. Barr, R.G. (1990a). The early crying paradox: A modest proposal. Human Nature, 1, 355–389. Barr, R.G. (1990b). The normal crying curve: What do we really know? Developmental Medicine and Child Neurology, 32, 356–362. Barr, R.G. (1996). Colic. In W.A. Walker, P.R. Durie, J.R.Hamilton, J.A.WalkerSmith, & J.B.Watkins (Eds.), Pediatric gastrointestinal disease: Pathophysiology, diagnosis, and management (2nd ed., pp. 241–250). St. Louis: Mosby. Barr, R.G. (1997). Excessive crying. In M.Lewis & A.J.Sameroff (Eds.), Handbook of developmental psychopathology (2nd ed.). New York: Plenum. Barr, R.G. (1998). Reflections on measuring pain in infants: Dissociation in responsive systems and “honest signalling.” Archives of Disease in Childhood: Fetal Neonatal Edition, 79, F152–F156. Barr, R.G., Chen, S., Hopkins, B., & Westra, T. (1996). Crying patterns in preterm infants. Developmental Medicine & Child Neurology, 38, 345–355. Barr, R. G., & Geertsma, M.A. (1993). Colic: The pain perplex. In N.L.Schechter, C. B.Berde, & M.Yaster (Eds.), Pain in infants, children, and adolescents (pp. 587–596). Philadelphia: Williams & Wilkins. Barr, R.G., Kramer, M.S., Leduc, D.G., Boisjoly, C., McVey-White, L., & Pless, I. B. (1988). Parental diary of infant cry and fuss behaviour. Archives of Diseases in Childhood, 63, 380–387. Barr, R.G., Kramer, M.S., Pless, I.B., Boisjoly, C., & Leduc, D.G. (1989). Feeding and temperament as determinants of early infant cry/fuss behaviour. Pediatrics, 84, 514– 521. Barr, R.G., McMullan, S.J., Spiess, H., Leduc, D.J., Yaremko, J., Barfield, R., Francoeur, T.E., & Hunziker, U.A. (1991). Carrying as colic “therapy”: A randomized controlled trial. Pediatrics, 87, 623–630. Barr, R.G., Rotmans, A., Yaremko, J., Leduc, D., & Francoer, T.E. (1992). The crying of infants with colic: A controlled empirical description. Pediatrics, 90, 14–21. Barr, R.G., Wooldridge, J.A., & Hanley, J. (1991). Effects of formula change on intestinal hydrogen production and crying and fussing behavior. Journal of Development and Behavioral Pediatrics, 12, 248–253. Bates, J.E., & Bayles, K. (1984). Objective and subjective components in mothers’ perceptions of their children from age 6 months to 3 years. Merrill-Palmer Quarterly, 30, 111–129.
Behavioral and physiological responsivity 87 Boyce, W.T., Barr, R.G., & Zeltzer, L.K. (1992). Temperament and the psychobiology of childhood stress. Pediatrics, 90, 483–486. Brazelton, T.B. (1984). Neonatal Behavioral Assessment Scale (2nd ed.). London: Spastics International Medical Publication. Carey, W.B. (1972). Clinical applications of infant temperament measures. Behavioral Pediatrics, 81, 823–828. Carey, W.B. (1984). “Colic”—Primary excessive crying as an infant-environment interaction. Pediatric Clinics of North America, 31, 993–1005. Coons, S. (1987). Development of sleep and wakefulness during the first six months of life. In C.Guilleminault (Ed.), Sleep and its disorders in children (pp. 17–27). New York: Raven Press. Coons, S., & Guilleminault, C. (1982). Development of sleep-wake patterns and nonrapid eye movement sleep stages during the first six months of life in normal infants. Pediatrics, 69, 793–798. Delta-Biometrics, Inc. (1994). Vagal Tone Monitor-II [Manual]. Bethesda, MD: Biomedical Technology. Forsyth, B.W.C., & Canny, P.F. (1991). Perceptions of vulnerability 3½ years after problems of feeding and crying behavior in early infancy. Pediatrics, 88, 757–763. Goldsmith, H.H., Buss, A.H., Plomin, R., Rothbart, M.K., Thomas, A., Chess, S., Hinde, R.A., & McCall, R.B. (1987). Roundtable: What is temperament? Four approaches. Child Development, 58, 505–529. Gormally, S.M., & Barr, R.G. (1997). Of clinical pies and clinical clues: Proposal for a clinical approach to complaints of early crying and colic. Good Practice Guide, 3, 137–153. Gunnar, M.R. (Ed.). (1990). The psychobiology of infant temperament. Hillsdale, NJ: Erlbaum. Gunnar, M.R. (1992). Reactivity of the hypothalamic-pitutiary-adrenocortical system to stressors in normal infants and children. Pediatrics, 90, 491–497. Gunnar, M.R., Brodersen, L., Krueger, K., & Rigatuso, J. (1996). Dampening of adrenocortical responses during infancy: Normative changes and individual differences. Child Development, 67, 877–889. Hill, D.J., Menahem, S., Hudson, I., Sheffield, L., Oberklaid, F., & Hosking, C.S. (1992). Charting infant distress: An aid to defining colic. Journal of Pediatrics, 121, 755–758. Hunziker, U.A., & Barr, R.G. (1986). Increased carrying reduces infant crying: A randomized controlled trial. Pediatrics, 77, 641–648. Larson, M.C. (1996). Dampening of stress responsivity in infancy, the developing hypothalamic-pituitary-adrenocortical circadian rhythm, and sleep. Unpublished dissertation, University of Minnesota, Minneapolis. Larson, M.C., White, B.P., Cochran, A., Donzella, B., & Gunnar, M.R. (1998). Dampening of the cortisol response to handling at 3 months in human infants and its relation to sleep, circadian cortisol activity, and behavioral distress. Developmental Psychobiology, 33, 327–337. Lehtonen, L. (1994). Infantile colic. Turku, Finland: Turun Yliopisto, Turku University Library. Lehtonen, L., Gormally, S., & Barr, R.G. (2000). Clinical pies for etiology and outcome
Annual progress in child psychiatry and child development 2000–2001
88
in infants presenting with early increased crying. In R.G.Barr, B.Hopkins, & J.A.Green (Eds.), Crying as a sign, a symptom, and a signal: Clinical, emotional, and developmental aspects of infant and toddler crying (pp. 67–95). London: MacKeith. Lehtonen, L., Korhonen, T., & Korvenranta, H. (1994). Temperament and sleeping pattern in colicky infants during the first year of life. Journal of Developmental and Behavioral Pediatrics, 15, 416–420. Lester, B.M., Boukydis, C.F.Z., Garcia-Coll, C.T., Hole, W.T., & Peucker, M. (1992). Infantile colic: Acoustic cry characteristics, maternal perceptions of cry, and temperament. Infant Behavioral Development, 15, 15–26. Lewis, M. (1989). Culture and biology: The role of temperament. In P.R.Zelazo & R. G.Barr (Eds.), Challenges to developmental paradigms: Implications for theory, assessment, and treatment (pp. 203–226). Hillsdale, NJ: Erlbaum. Littman, B., & Parmelee, A.H.J. (1978). Medical correlates of infant development. Pediatrics, 61, 470–474. McMillen, I.C., Kok, J.S.M., Adamson, T.M., Deayton, J.M., & Nowak, R. (1991). Development of circadian sleep wake rhythms in preterm and full-term infants. Pediatric Research, 29, 381–384. Miller, A.R., & Barr, R.G. (1991). Infantile colic: Is it a gut issue? Pediatric Clinics of North America, 38, 1407–1423. Miller, A.R., Barr, R.G., & Eaton, W.O. (1993). Crying and motor behavior of sixweekold infants and postpartum maternal mood. Pediatrics, 92, 551–558. Moore-Ede, M.C. (1986). Physiology of the circadian timing system: Predictive versus reactive homeostasis. American Journal of Physiology, 250, 735–752. Norusis, M.J., & SPSS, Inc. (1994). SPSS Advanced Statistics 6.1 [Computer software]. Chicago: SPSS Inc. Papousek, M., & von Hofacker, N. (1995). Persistent crying and parenting: Search for a butterfly in a dynamic system. Early Development and Parenting, 4, 209–224. Papousek, M., & von Hofacker, N. (1998). Persistent crying in early infancy: A nontrivial condition of risk for the developing mother-infant relationship. Child Care, Health and Development, 24, 395–424. Porges, S.W. (1985). Method and apparatus for evaluating rhythmic oscillations in aperiodic physiological response systems. U.S. Patent No. 4,510,944. http://patents.uspto.gov/cgi-bin/ (21 July 2000). Porges, S.W. (1995). Cardiac vagal tone: A physiological index of stress. Neuroscience and Biobehavioral Reviews, 19, 225–233. Raiha, H., Lehtonen, L., Korhonen, T., & Korvenranta, H. (1997). Family functioning 3 years after infantile colic. Journal of Developmental and Behavioral Pediatrics, 18, 290–294. Rao, M., Blass, E.M., Brignol, M.J., Marino, L., & Glass, L. (1997). Reduced heat loss following sucrose ingestion in premature and normal human newborns. Early Human Development, 48, 109–116. Rautava, P., Lehtonen, L., Helenius, H., & Silanpaa, M. (1995). Infantile colic: Child and family three years later. Pediatrics, 96, 43–47. Riese, M.L. (1995). Mothers’ ratings of infant temperament: Relation to neonatal latency to soothe by pacifier. The Journal of Genetic Psychology, 156, 23–31.
Behavioral and physiological responsivity 89 Ross Laboratories. (1992). Growth and development charts for infants. Columbus, OH. Rothbart, M.K. (1981). Measurement of temperament in infancy. Child Development, 52, 569–578. Rothbart, M.K. (1989). Biological processes in temperament. In G.A.Kohnstamm, J. E.Bates, & M.K.Rothbart (Eds.), Temperament in childhood (pp. 77–103). New York: Wiley. Rothbart, M.K., & Derryberry, D. (1981). Development in individual differences in temperament. In M.E. Lamb & A.L.Brown (Eds.), Advances in developmental psychology (Vol. 1, pp. 37–85). Hillsdale, NJ: Erlbaum. Sauls, H.S., & Redfern, D.E. (Eds.). (1997). Colic and excessive crying. Columbus, OH: Ross Products Division Abbott Laboratories. Smyth, J.M., Ockenfels, M.C., Gorin, A.A., Catley, D., Porter, L.S., Kirschbaum, C., Hellhammer, D.H., & Stone, A.A. (1997). Individual differences in the diurnal cycle of cortisol. Psychoneuroendocrinology, 22, 89–106. Stifter, C., & Bono, M. (1998). The effect of excessive infant crying on maternal selfperceptions and mother-infant attachment. Child: Care, Health and Development, 24, 339–351. Stifter, C.A., & Braungart, J. (1992). Infant colic: A transient condition with no apparent effects. Journal of Applied Developmental Psychology, 13, 447–462. St. James-Roberts, I., Conroy, S., & Hurry, J. (1997). Links between infant crying and sleep-waking at six weeks of age. Early Human Development, 48, 43–52. St. James-Roberts, I., Conroy, S., & Wilsher, K. (1995). Clinical, developmental and social aspects of infant crying and colic. Early Development and Parenting, 4, 177– 189. St. James-Roberts, I., Conroy, S., & Wilsher, K. (1998). Links between maternal care and persistent infant crying in the early months. Child: Care, Health and Development, 24, 353–376. St. James-Roberts, I., & Halil, T. (1991). Infant crying patterns in the first year: Normal community and clinical findings. Journal of Child Psychology and Psychiatry, 32, 951–968. St. James-Roberts, I., Hurry, J., & Bowyer, J. (1993). Objective confirmation of crying durations in infants referred for excessive crying. Archives of Disease in Childhood, 68, 82–84. St. James-Roberts, I., & Plewis, I. (1996). Individual differences, daily fluctuations, and developmental changes in amounts of infant waking, fussing, crying, feeding and sleeping. Child Development, 67, 2527–2540. The Observer 3.0 [Computer software]. 1993. Wageningen, The Netherlands: Noldus Information Technology. Base package for DOS. Thomas, A., Chess, S., & Birch, H. (1968). Temperament and behavior disorders in children. New York: New York University Press. Thompson, R.A. (1994). Emotion regulation: A theme in search of definition. In N. A.Fox (Ed.), The development of emotion regulation: Biological and behavioral considerations (pp. 25–52). Chicago: University of Chicago Press. Treem, W.R. (1994). Infant colic: A pediatric gastroenterologists’s perspective. Pediatric Clinics of North America, 41, 1121–1138.
Annual progress in child psychiatry and child development 2000–2001
90
Weissbluth, M. (1984). Crybabies. Coping with colic: What to do when baby won’t stop crying. New York: Priam. Weissbluth, M., Christoffel, K.K., & Davis, T. (1984). Treatment of infantile colic with dicyclomine hydrochloride. Journal of Pediatrics, 104, 951–955. Wessel, M.A., Cobb, J.C., Jackson, E.B., Harris, G.S., & Detwiler, A.C. (1954). Paroxysmal fussing in infancy, sometimes called “colic.” Pediatrics, 14, 421–434. Wolke, D., Meyer, R., Ohrt, B., & Riegel, K. (1994). Prevalence and persistence of sleeping problems during the pre-school years: A prospective investigation in a representative sample of South German children. Praxis der Kinderpsychologie und Kinderpsychiatrie, 43, 331–339. Wolke, D., Meyer, R., Ohrt, B., & Riegel, K. (1995). The incidence of sleeping problems in preterm and fullterm infants discharged from neonatal special care units: An epidemiological longitudinal study. Journal of Child Psychology and Psychiatry and Allied Disciplines, 36, 203–223.
PART I: DEVELOPMENTAL ISSUES
6 Imaginary Companions of Preschool Children Tracy R.Gleason, Anne M.Sebanc, and Willard W.Hartup
The developmental significance of preschool children’s imaginary companions was examined. Mothers of 78 children were interviewed about their children’s social environments and imaginary companions (if their children had them). Results revealed differences between invisible companions and personified objects (e.g., stuffed animals or dolls) in terms of the pretend friends’ stability and ubiquity, identity, and relationship with the child. Relationships with invisible companions were mostly described as sociable and friendly, whereas personified objects were usually nurtured. Mothers reported that personification of objects frequently occurred as a result of acquiring a toy, whereas invisible friends were often viewed as fulfilling a need for a relationship. Compared to children without imaginary companions, children with imaginary companions were more likely to be firstborn and only children.
INTRODUCTION Studies of the imaginary companions of preschool children are widely scattered in the developmental and psychoanalytic literature. Most of these investigations attempt to explain why some children create pretend friends and others do not. In particular, the structure of a child’s social environment has been emphasized. Results, however, are inconsistent and largely ineffective in explaining the origins of the phenomenon. Furthermore, identification of the developmental significance of imaginary companions has been compromised by a paucity of descriptive studies of pretend friends as well as by variations among researchers’ definitions. In this investigation, we attempt to shed light on the role that imaginary companions play in children’s lives and on the reasons why some children have these charming associates. We begin by reviewing issues of definition and categorization. Subsequently, we review the correlates of having an imaginary companion that have been suggested by previous research and that prompted our examination of conditions associated with the phenomenon.
Annual progress in child psychiatry and child development 2000–2001
92
DEFINING AND CATEGORIZING IMAGINARY COMPANIONS Many authors who have studied imaginary companions (Manosevitz, Fling, & Prentice, 1977; Manosevitz, Prentice, & Wilson, 1973; Meyer & Tuber, 1989; Somers & Yawkey, 1984) used a definition similar to that provided by Svendsen (1934): an invisible character, named and referred to in conversation with other persons or played with directly for a period of time, at least several months, having an air of reality for the child, but no apparent objective basis. This excludes that type of imaginative play in which an object is personified, or in which the child himself assumes the role of some person in his environment, (p. 988) Other researchers (Jersild, Markey, & Jersild, 1933; Mauro, 1991; D.Singer & Singer, 1990; J.Singer & Singer, 1981) have included in their definitions objects that children personify and animate, sometimes distinguishing them from invisible companions and sometimes not. Well-known examples of such personified objects include Winnie the Pooh (Milne, 1924) and Hobbes, from the comic strip Calvin and Hobbes. Role-playing, in which the child impersonates a specific character every day for a long period of time, is also occasionally discussed in the imaginary companion literature as an activity similar to having an invisible imaginary companion (Sperling, 1954). Ames and Learned (1946), for example, studied a group of children who impersonated characters, had invisible companions or personified objects, and engaged in imaginative play more generally. Ames and Learned analyzed their sample first as a whole and then subsequently by activity type, thereby acknowledging that although these phenomena were “related” (p. 147), there were important distinctions among them. As one would expect, definitional differences are reflected in the incidences of imaginary companions reported. When multiple forms of imaginary companions are included in the definition, incidence rises. Estimates of the proportion of children who create imaginary companions range from 6% (Harvey, 1918) to 65% (D.Singer & Singer, 1990). Whether children who create invisible companions, personify objects, or impersonate characters are included in a study often depends on the research agenda of the investigator. Some researchers include only those with invisible companions because of a primary interest in mental imagery (Harvey, 1918; J. Singer & Streiner, 1966). In contrast, Taylor and Carlson (1997), in an investigation of the relation between fantasy and theory of mind, classified children who fit at least one of these three categories as a “high fantasy” group of chil-dren for comparison with a “low fantasy” group of children. This connection between definition and content in empirical studies emphasizes the fact that research on imaginary companions has frequently emerged secondarily from investigations in other areas, such as social competence (Harter & Chao, 1992), creativity (Schaefer, 1969), and language (Fulmer, 1995). The developmental significance of imaginary companions has been obscured by the fact that relatively few investigators have set out to examine imaginary companions for
Imaginary companions of preschool children 93 their own sake (Taylor, 1999, is a notable exception). Little is known about the appearance of an imaginary companion in a child’s life, although some descriptive information has been obtained on existing companions. For example, more than half of all imaginary companions have no apparent trigger (Masih, 1978), and children are rarely able to explain their companions’ appearance (Mauro, 1991; Svendsen, 1934). Parents have reported that some invisible friends are imaginary extensions of real people (especially those the child admires), of names the child heard once, or of characters from stories, movies, or television (Ames & Learned, 1946; Harvey, 1918; Hurlock & Burstein, 1932; Manosevitz et al., 1973; J.Singer & Singer, 1981; Svendsen, 1934; Vostrovsky, 1895). Invisible companions are most frequently human, but children also create imaginary animals, aliens, and monsters (Ames & Learned, 1946; Benson & Pryor, 1973; Harvey, 1918; Vostrovsky, 1895). They appear singly and in groups, but which is more common is unclear (Manosevitz et al., 1973; Svendsen, 1934; Taylor, Cartwright, & Carlson, 1993; Vostrovsky, 1895). Regardless of form, imaginary companions of all types are most frequently children, although their exact ages are rarely specified (Harvey, 1918; Hurlock & Burstein, 1932; Masih, 1978; Vostrovsky, 1895). As for gender, the majority of children’s imaginary companions are male, given that boys almost always create same-sex companions and girls create female companions only slightly more often than they create male companions (Jersild et al., 1933; Manosevitz et al., 1973; D.Singer & Singer, 1990; J.Singer & Singer, 1981). Speculation regarding the origins of personified objects and extensive roleplaying frequently refers to activities common to most young children. For example, stuffed animals, dolls, and blankets are often animated the moment they are adopted by a child and may evolve into full-blown personified objects over time. Alternatively, personified objects may originate as what Winnicott (1953) termed transitional objects, concrete representations of an external source of comfort such as the mother. As a child’s need for a security object decreases with increasing maturity, a transitional object may become imbued with personality and agency and emerge as a personified object (D.Singer & Singer, 1990). Why such a metamorphosis occurs for some transitional objects but not others is unknown. Similarly, consistent and extensive role-playing akin to having an imaginary companion probably emerges from the periodic dramatic play characteristic of most children. Again, the reasons why only some children enhance their role-playing in this manner are unclear. Although researchers studying imaginary companions frequently provide details such as the species or physical characteristics of these creations, the manner in which pretend friends fit into children’s lives has largely been uninvestigated. For example, few researchers have addressed whether imaginary companions tend to be constant or episodic friends or whether their presence is associated with specific contexts, such as a room in a child’s home or traveling in the car. Whether imaginary companions are playmates, advisors, or in need of caretaking (or all three!) has not been well established, nor has their prevalence in children’s lives. In fact, whether different children treat their imaginary companions similarly has not been explored. In essence, the quality of children’s relationships with imaginary companions has been neglected in favor of investigating simply whether children have them. In light of the incomplete descriptive information available on imaginary companions, as well as the pancity of data on
Annual progress in child psychiatry and child development 2000–2001
94
children’s relationships with pretend friends, our first goal in this study was to examine the range of characteristics of imaginary companions, especially their modes of “interaction” with children. Examination of the stability, ubiquity, and identities of imaginary companions, as well as the qualities of child-imaginary companion relationships, was a primary research goal rather than a purpose secondary to some other line of inquiry. We expected such an approach to provide a more complete and thorough rendering of imaginary companions than is typically found in the literature. As for definitions, although the various types of imaginary companions are sometimes linked in research, little evidence to date has addressed whether the three varieties of imaginative activity should logically be subsumed by the overarching category of “imaginary companion.” A notable exception is the work of Taylor and Carlson (1997), who demonstrated that 4-year-old children who engage in extensive role-playing or who have imaginary companions of either type resemble each other and differ from their peers in performance on theory-of-mind tasks. In hopes of distinguishing between types of imaginary companion with respect to relationships, therefore, we made two decisions regarding the definition of imaginary companion: (a) We limited our definition to the two forms that most directly provide for quasisocial interaction, invisible friends and personified objects, and (b) in the course of data collection we distinguished between these two types rather than treating them as identical entities.
CORRELATES OF HAVING AN IMAGINARY COMPANION One of the most common notions concerning imaginary companions is that they may compensate for a child’s poor social relationships or loneliness (Bender & Vogel, 1941; Harvey, 1918; Manosevitz et al., 1973). Some empirical evidence has indicated otherwise, however, suggesting that children with pretend friends are particularly sociable by nature (Mauro, 1991; Partington & Grant, 1984; D.Singer & Singer, 1990). Taylor (1999) also asserted that imaginary friends are primarily a source of companionship for children. Significantly, the only connection reported between the occurrence of imaginary companions and the child’s social environment concerns siblings. Children with imaginary friends are more likely to be firstborn and only children than later-born children (Ames & Learned, 1946; Masih, 1978; D.Singer & Singer, 1990; J.Singer & Singer, 1981; Svendsen, 1934). Children who create imaginary companions and who also have siblings spend less time playing with their siblings (Masih, 1978) and are further apart in age from their siblings than are children without imaginary companions (Manosevitz et al., 1973; D.Singer & Singer, 1990). Nevertheless, a few investigators have not found statistically significant connections between birth order or number of siblings and the formation of an imaginary companion (Hurlock & Burstein, 1932; Manosevitz et al., 1973). In addition, children missing a parent because of divorce or death are not more likely to create imaginary companions than are children from twoparent families, nor does the presence of an imaginary companion relate to number of household members, pets, or playmates or to the amount of time a child spends with playmates (Manosevitz et al., 1973; Masih, 1978). Given that some of the evidence connecting the presence of imaginary companions to
Imaginary companions of preschool children 95 children’s social environments is inconsistent or has not been replicated, our second goal was to investigate the relation between the creation of imaginary companions and children’s social surroundings. We included aspects of children’s social environments both within the family, such as birth order and number of siblings, and outside the family, such as preschool experience and friends (as opposed to playmates).
METHOD Participants Participants were the mothers1 of 78 preschool children (mean age=49 months; range=31 to 73 months) attending two university-affiliated preschools. The children included 45 girls and 33 boys from middle-class and upper middleclass backgrounds. The sample was 78% white, 1% African American, 9% Asian, 1% Hispanic, and 10% other (or not identified). Ninety-one percent of the mothers were married, 6% were single, and 3% were divorced or separated. Children were classified into three groups according to whether an imaginary companion was evident (as explained in the following) and its type: (a) children with an invisible imaginary companion (n=24; 12 girls and 12 boys); (b) children with a personified object (n=21; 15 girls and six boys); and (c) children without an imaginary companion (n=33; 18 girls and 15 boys). According to their mothers, eight children had both invisible companions and personified objects; although data were gathered on both companions, only the data from the invisible companions were used unless otherwise noted. Some concern regarding the accuracy of mothers’ reports on children’s imaginary companions has been expressed in the literature (e.g., Taylor, 1999). These concerns are threefold. First, children do not always appear to share their imaginary companions with their mothers. According to Schmechel (1975, cited in Taylor et al., 1993), approximately one-third of parents are unfamiliar with their children’s imaginary companions. Parent reports have also been questioned because invisible companions, in particular, are sometimes described differently by children from one time to the next. The third concern regarding maternal report as a source for data is that mothers’ reports may be biased according to their interpretations of their children’s behavior. Moreover, such bias may be systematic with regard to whether the companion is invisible or a personified object. These limitations should be taken into account when interpreting the findings of this research. 1
Three fathers also participated in the interviews.
Annual progress in child psychiatry and child development 2000–2001
96
PROCEDURE AND MATERIALS Recruitment Mothers of all children in the two preschools received a letter explaining the purpose of the study and subsequently were contacted by telephone. If a mother indicated a willingness to participate, the experimenter provided her with Svendsen’s (1934) definition of an invisible imaginary companion and asked her if her child had such a friend. If so, the researcher asked for a description of the friend to be certain the mother had understood the question correctly. If the mother reported the child did not appear to have an invisible companion, the experimenter asked about personified objects. Specifically, the mother was told that many children have an object of which they are fond, such as a stuffed animal or a doll, which they seem to treat as animate. Examples such as Winnie the Pooh and Hobbes were mentioned, and the mother was asked if her child had a similar friend with whom he or she interacted regularly. Imaginary companions had to be present in the child’s life for a minimum of 1 month, according to mothers’ reports, to be included in the study. Mothers of children whose imaginary companions had been present for less than 1 month were contacted again after a few weeks and were then either recruited if the companion was still around or dropped from the study if it had disappeared. Mothers of target children (those with an invisible friend or a personified object present for at least 1 month) were invited to participate in the interview immediately. Mothers of control group children (those with no imaginary companion of any kind) were also asked for an interview if their child matched a child in the target group on age and gender. Interview At the time of the interview, mothers signed a consent form to participate and filled out a brief form asking for background information. A semistructured interview was then conducted to provide descriptive information on the invisible and personified-object imaginary companions of the preschool children and on the children themselves. Mothers of children with imaginary companions answered questions regarding these friends and their origins. All mothers answered questions regarding several areas postulated to be related to the formation of imaginary companions, including elements of their children’s social environments. Imaginary Companions. Mothers of children with invisible friends or personified objects were asked a series of questions from three categories. Variables concerning the mother’s perception of the companion’s stability and ubiquity included the number of months the companion had existed, the mother’s mode of discovery of the companion (observation, child told mother, or a combination of the two), the timing of the companion’s appearance (sudden or gradual), the diversity of the situations in which the companion appeared (low, medium, or high), how well-known the companion was (to
Imaginary companions of preschool children 97 immediate family only, family and nonfamilial adults, everyone), and whether the companion participated in the child’s daily routines, traveled with the child, attended school, or had undergone any sort of change or metamorphosis (e.g., switching gender). Variables concerning the mother’s perception of the companion’s identity included configuration (single or multiple), the specific number of companions (if countable), and the companion’s manifestation (human, animal, or other). Mothers also specified whether the companion had superpowers, demonstrated preferences, and had an age, either specific or general (e.g., a child). Mothers of children with invisible companions were asked if the pretend friends had physical characteristics and, if so, whether these characteristics had developed suddenly or gradually. Last, mothers were asked how they knew about the companion’s characteristics. Specifically, mothers reported whether the companion’s characteristics (physical or otherwise) resembled the child’s or related to elements of the child’s life. For example, a mother might have learned about a companion’s hair color when brushing her child’s hair or about the companion’s affinity for a certain food during a meal. Interview questions also addressed qualities of the child-imaginary companion relationship, such as orientation (horizontal or vertical), presence or absence of caregiving, conflict, or didactic instruction (child as teacher), and whether the companion received discipline. Mothers reported on children’s involvement of imaginary companions in their own misbehavior, on whether the companions were ever blamed for misconduct, and on whether children provided their companions with guidance or praised them for good behavior. Whether the companion participated in role-playing and had conversations with the child was also investigated. Last, mothers reported whether the child made requests of the parent on behalf of the imaginary companion. These last questions were included because of anecdotal evidence demonstrating that parents are frequently asked to behave as if imaginary companions are real by providing space or food for them. Mothers’ Explanations of Children’s Creation of Imaginary Companions. Mothers of children with imaginary companions were asked an open-ended question about why they believed their children had created these friends. Responses were divided into 13 categories (present or absent), which sorted loosely into three groups of explanations: (a) those involving the child’s relationships (birth of a sibling, family change, lack of playmates, need for a particular kind of relationship, and birth order); (b) those involving stressors (moving, stress related to school, and medical conditions); and (c) those attributing the creation of an imaginary companion to everyday experiences. Companions described by the last group of explanations were based on typical experiences that were nevertheless remarkable to the child. Categories in this group included common events (e.g., an encounter with a neighborhood cat prompted creation of an invisible one), incarnations of fictional characters, and what mothers termed “normal development.” Finally, invisible friends were coded as to whether or not the companion represented a real person, and personified objects were coded as to whether or not personification began with the acquisition of the object. Social Environment. Mothers reported on the child’s ethnicity, age in months, and gender, as well as on properties of the child’s social environment, both within and outside the family. Information on the social environment within the family included the
Annual progress in child psychiatry and child development 2000–2001
98
mother’s marital status, the target child’s birth order, and the number of siblings. Features of the child’s social environment outside the family included the number of years the child had been in preschool (1 versus 2 or more), the child’s reaction to preschool (4point scale ranging from enjoys to doesn’t like at all), the presence or absence of friendships (either in or out of preschool), and the source of the information on friendships (from the child, teacher, or observation). Reliability. Kappas were calculated on 26 cases (33%) to measure the agreement between two coders on all variables coded from the interview. All kappas were significant and >0.72.
RESULTS The results are reported in two sections corresponding to the goals of the project. Issues related to defining and categorizing invisible friends and personified objects are addressed first, including similarities and differences between the two types of imaginary companions with respect to mothers’ descriptions of stability and ubiquity, companion identities, and qualities of the child-imaginary companion relationship. Mothers’ explanations for their children’s imaginary creations are also discussed in this section. Next we consider the potential correlates of having an imaginary companion, using comparisons among the three groups of children. Defining and Categorizing Imaginary Companions Incidences of invisible imaginary companions and personified objects were approximately 19% and 23%, respectively. These values were calculated by dividing the number of children in each category by the total number of children whose mothers were invited to participate (N=128) regardless of whether or not they were interviewed. These estimates are most likely low given that not all children share their imaginary companions with their parents. Table 6.1 provides some examples of both types of imaginary companion described by mothers.
TABLE 6.1 Examples of Imaginary Companions Described by Mothers
Name
Description
Invisible friends Star Friends and Groups of preschool-aged human friends with whom the child Heart Fan Club had birthdays, went to the fair, and spoke a language called Hobotchi Goofy and Disney character Goofy and invented relatives including Goofy company Jr. (son), Max (cousin), and the Doctor, who tended Goofy’s asthma (child had asthma as well) Herd of cows Cows of many colors and varying sizes who were often fed or diapered like infants. Discovered when the child’s father
Imaginary companions of preschool children 99
Danny Maybe Blankie Bandit Franklin Dirtybaby Percy and Gordon
accidentally stepped on one An imaginary version of the child’s best friend from preschool A human of varying gender whom the child routinely summoned by shouting out the front door of the family’s house Personified objects A baby blanket that was sometimes an infant or a pet and also served as a cape that endowed the child with superpowers A stuffed raccoon that was a constant companion and did what the child did—including getting her throat checked at preschool A brown purse of the mother’s that the child adopted as a puppy. When it fell apart and was replaced, the new purse became the new Franklin. A stuffed animal (species no longer clear) that was thrown into the classroom upon the child’s arrival at school and observed for harm before the child would enter Trains from the television show Shining Time Station who accompanied the child throughout the day; Gordon was occasionally “crabby.” Invisible Friends Versus Personified Objects
Stability and Ubiquity. According to maternal report, personified objects were present longer on average (M=17.8 months, SD=12.7 months) than were invisible friends (M=9.8 months, SD=7.9 months), t(27) = −2.37, p< 0.01. Table 6.2 lists the variables associated with stability and ubiquity on which we compared personified objects and imaginary companions using chisquare analyses. For all but two variables, mothers reported personified objects as significantly more stable and ubiquitous than invisible companions. The two types of imaginary companions did not differ significantly on timing of appearance because both invisible companions and personified objects tended to appear suddenly rather than gradually. Companion types also did not differ according to whether they had undergone a change; approximately 40% of all imaginary companions went through some sort of major change over the course of their existence (e.g., switching gender or dying). Companion Identities. Mothers reported that children who created invisible friends were more likely to have multiple companions than were children with personified objects, who typically had single companions (see Table 6.3). Multiple companions included distinct individuals (e.g., Goofy, Goofy Jr., Max, and the Doctor; see Table 6.1) or undifferentiated groups (e.g., the Heart Fan Club; a herd of cows). Not surprisingly, the number of companions created by children with invisible friends (M=2.4, SD=1.9) was also significantly higher than the number of companions created by children with personified objects (M=1.2, SD=0.4), t(17.2)=2.60, p<0.05. (The means and standard deviations for number of companions exclude children who had uncountable imaginary companions.) Invisible friends were most often human, and personified objects were most frequently animals.
Annual progress in child psychiatry and child development 2000–2001
100
TABLE 6.2 Stability and Ubiquity of Invisible Friends and Personified Objects: Chi-Square Analyses
Invisible Friends Personified Objects Variable df N χ2 (%) (%) Discovery 2 44 15.31** Observation of child 13.0 57.1 Child told parent 82.6 23.8 Both 4.4 19.1 Timing of appearance 1 45 .92 Sudden 70.8 57.1 Gradual 29.2 42.9 2 45 10.81** Situational diversity Low 45.8 4.8 Medium 37.5 47.6 High 16.7 47.6 Who knows companion 2 44 9.05* 26.1 9.5 Family only 47.8 19.0 Family and other adults 26.1 71.4 Everyone 1 45 14.32** Participation in daily routines Yes 50.0 100.0 No 50.0 0.0 Travels with child 1 44 6.38* Yes 56.5 90.5 No 43.5 9.5 Attends school 1 45 9.64** Yes 20.8 66.7 No 79.2 33.3 1 41 0.09 Has undergone a change Yes 43.5 38.9 No 56.5 61.1 Note. For invisible friends, n=24; for personified objects; n=21. *p≤0.01, **p≤0.005. Invisible companions and personified objects differed in terms of some of the characteristics typically attributed to them (see Table 6.3). Statistically equivalent proportions of invisible friends and personified objects had superpowers (approximately 30%) and an approximate age (75%), but a marginally greater proportion of personified objects than invisible companions had preferences that were specified by the children. Personified objects had identifiable physical characteristics, of course, but only 75% of mothers of children with invisible friends could provide physical descriptions for the companions. For the invisible friends who had known physical characteristics, 68% of mothers reported that the characteristics were acquired over a long period of time and
Imaginary companions of preschool children 101
TABLE 6.3 Identities of Invisible Friends and Personified Objects: Chi-Square Analyses*
Invisible Friends (%)
Personified Objects (%)
Variable df N Configuration 1 45 37.5 71.4 Single Multiple 62.5 28.6 Manifestation 2 56** Human 51.5 17.4 Animal 36.4 65.2 Other 12.1 17.4 Superpowers 1 44 Yes 39.1 23.8 No 60.9 76.2 Preferences 1 44 Yes 60.9 85.7 No 39.1 14.3 1 36 Age Yes 75.0 75.0 No 25.0 25.0 Child-imaginary companion 1 41 resemblance Yes 71.4 65.0 No 28.6 35.0 Characteristics relate to child’s 1 32 life Yes 70.0 33.3 No 30.0 66.7 Note. For invisible friends, n=24; for personified objects, n=21. **Each imaginary companion (or homogenous group) was coded for this variable, so the N reflects the presence of multiple companions of varied manifestations. **p≤0.05, ***p≤0.08.
χ2 5.18** 6.81**
1.19 3.42*** 0.00 0.20
4.10**
32% reported that the characteristics seemed to be full-blown at first appearance. For both types of imaginary friends, the companions’ characteristics (physical or otherwise) were reported by mothers to resemble their children’s, but as depicted in Table 6.3, a higher proportion of mothers of children with invisible companions reported that these characteristics emerged through situations in the child’s life. Quality of the Child-Imaginary Companion Relationship. Differences between the relationships children developed with invisible companions and the relationships they developed with personified objects are shown in Table 6.4.
Annual progress in child psychiatry and child development 2000–2001
102
In general, relationships with personified objects were more frequently oriented vertically than horizontally in that the child fulfilled a nurturing, parentlike role for the companion, whereas relationships with invisible companions were egalitarian, like friendships. Moreover, as displayed in the table, differences between these two types of relationships extended to behavioral domains, such that imaginary companions were distinguished by the presence or absence of some parenting activities within the relationship, including caregiving, providing guidance, and praising companions for good behavior. In addition, compared to children with invisible companions, greater proportions of children with personified objects tended to talk with their imaginary companions and include them in role-playing.
TABLE 6.4 Child-Imaginary Companion Relationship Qualities: Chi-Square Analyses
Variable Relationship orientation Horizontal Vertical, child in charge Vertical, companion in charge Caregiving Yes No Conflict Yes No Didactic instruction Yes No Discipline Yes No Involved in misbehavior Yes No Blame Yes No Guidance Yes No Praise
Invisible Friends (%)
Personified Objects (%)
57.1 28.6 14.3
20.0 75.0 5.0
21.0 79.0
81.0 19.0
8.3 91.7
4.8 95.2
25.0 75.0
42.9 57.1
26.1 73.9
38.1 61.9
43.5 56.5
42.9 57.1
30.4 69.6
28.6 71.4
27.1 72.9
df N 2 41*
χ2 8.84**
1 45
16.20***
1 45
0.23
1 45
1.61
1 44
.73
1 44
0.00
1 44
.02
2 44
5.74†
3 45
6.92‡
61.9 38.1
Imaginary companions of preschool children 103
Yes 17.0 52.0 No 83.0 48.0 1 44 3.20‡ Participates in roleplaying Yes 26.1 52.4 No 73.9 47.6 Has conversations 1 45 4.75† Yes 62.5 90.5 No 37.5 9.5 Makes requests of parent 1 45 2.67 Yes 37.5 61.9 No 62.5 38.1 Note. For invisible friends, n=24; for personified objects, n=21. *Four children (three with invisible friends, one with a personified object) were not included in this analysis because of differing relationship orientations with multiple imaginary companions. **p≤.01, *** p≤0.005, ‡p≤0.08, †p≤0.05. A few variables related to parentlike activities did not differentiate invisible companions and personified objects, perhaps because of low incidence. For example, the two target groups were not distinguished by whether the child provided the companion with didactic instruction or discipline, and similar proportions of children in each group blamed their companions for misconduct at least once and involved them in their own misbehavior (see Table 6.4). Finally, the proportion of mothers of children with personified objects who reported that their children asked them to fulfill their imaginary companions’ requests was statistically equivalent to the proportion of mothers of children with invisible friends who received similar appeals. Requests included such activities as buckling the imaginary companion’s seat belt in the car or providing a seat at the table. Mothers’ Explanations of Imaginary Companion Creation The number of explanations mothers provided for their children’s creation of imaginary companions ranged from zero to five, and mothers of children with invisible friends reported more reasons (M=2.21, SD=1.02) than mothers of children with personified objects (M=1.13, SD=0.86), t(52)=4.2, p= 0.000. (Mothers whose children had both an invisible imaginary companion and a personified object answered the question twice, once for each type of companion, and both answers were included in the analyses.) Greater proportions of invisible companions than of personified objects were explained by reasons related to relationships, but the two groups did not differ significantly according to any of the reasons related to stresses or everyday experiences (see Table 6.5). Finally, 21% of invisible companions were based on a real person, and mothers explained 50% of personified objects as resulting from acquisition of the object.
Annual progress in child psychiatry and child development 2000–2001
104
TABLE 6.5 Mothers’ Understanding of Why Their Children Created Imaginary Companions: Percentages Endorsing Designated Explanations
Explanation Group and Category
Invisible Friends (%)
Personified Objects (%)
Relationships 12.5 10.0 Birth of a sibling 29.2 10.0 Changes in the family 20.8 3.3 Lack of playmates Need for relationship 37.5 10.0 Birth order 29.2 3.3 Stresses Moving 4.2 3.3 School-related stress 8.3 16.7 Medical conditions 0.0 10.0 Everyday experiences Common events 20.8 10.0 Fictional characters 29.2 10.0 Normal development 29.2 26.7 Basis in real person 20.8 — Acquisition — 50.0 Note. For invisible friends, n=24; for personified objects, n=29. *p≤0.08, ** p≤0.05; *** p≤0.01.
χ2 (1, N=54)
0.08 3.25* 4.13** 5.83*** 7.05*** 0.03 0.82 2.54 1.24 3.25* 0.04 — —
Correlates of Having an Imaginary Companion Children with invisible friends, children with personified objects, and children without imaginary companions did not differ significantly according to ethnicity, and comparisons by age and gender were not conducted because these variables were used for matching purposes. Children with invisible friends, children with personified objects, and children without imaginary companions did not differ significantly according to number of years in preschool, reaction to preschool, presence or absence of friends, how mothers knew about their friends, or mothers’ marital status. Significant differences in birth order emerged, however, in that children with invisible friends were more likely than children in the other two groups to be firstborn or only children, χ2 (4, N=78)=14.23, p<0.007. Eighty-eight percent of children with invisible friends were firstborn or only children, compared with 67% of children with personified objects, and 45% of children without imaginary companions. None of the children who were middle-born (n=5; 6.4% of the total sample) had an imaginary companion. A one-way analysis of variance on number of siblings was significant, F(2, 75)=8.27, p=0.001. Post hoc tests revealed that children without imaginary companions had more siblings (M=1.18, SD=0.77) than did children with invisible companions (M=0.54, SD=0.51) or children with personified objects
Imaginary companions of preschool children 105 (M=0.62, SD=0.59). The two groups of children with imaginary companions did not differ significantly from each other in mean number of siblings.
DISCUSSION The findings presented in this chapter suggest that combining invisible imaginary companions and personified objects under the category of “imaginary companions” may not necessarily be appropriate when considering their functional significance in the preschool years. In particular, the differences inherent in the descriptions mothers provided suggest that the two types of imaginary companions may afford different types of relationships for children. In addition, differences between the groups of children examined in this study appear related to birth order and number of siblings. Distinguishing Invisible Friends and Personified Objects One of our goals in this project was to provide thorough descriptions of imaginary companions with an eye toward deciding if invisible friends and personified objects differed in characteristics other than physical presence. The fact that the two types of imaginary companions were distinguished by variables concerning mothers’ perceptions of their stability and ubiquity, identity, and relationship qualities implies that they are indeed distinct. In particular, personified objects appear to be more ubiquitous in children’s lives than invisible friends are, even though most children with invisible friends have more than one. Although the reasons why these distinctions exist are unclear, they suggest that not all children who create imaginary companions do so for the same reasons. These two types of imaginary companions should therefore be distinguished in discussions of their developmental significance. A word is in order with respect to the issue of maternal report as a basis for these findings. At first glance, the fact that personified objects appear more stable and ubiquitous than invisible friends seems pedantic. The ethereal nature of invisible friends, in contrast to the tangibility of personified objects, contributes to mothers’ perceptions of the former as less stable and ubiquitous. Mothers may also have overestimated the ubiquity of personified objects by confusing episodes in which the personified objects functioned as animated beings with episodes in which they were merely comforting objects. Conversely, the mere mention of an invisible companion makes it present (and potentially highly ubiquitous), and children with invisible friends in this study shared them with others. Consequently, differences between the two types of companions in terms of stability and ubiquity are probably not a function of mothers’ perceptions alone. The findings with respect to the qualities of the child-imaginary companion relationships provide the most persuasive evidence that personified objects and invisible friends differ in nature and function. Although not universal, the typically horizontal nature of children’s relationships with invisible companions is reminiscent of a friendship, whereas the vertical nature of relationships with personified objects is more akin to a parent-child relationship (where the child is the parent and the object is the child). One interpretation of these findings is that children are purposefully creating a
Annual progress in child psychiatry and child development 2000–2001
106
specific type of relationship. Of course, mothers observe children carrying their personified objects around, much as a parent totes a child, and this behavior may artificially inflate the percentage of children’s relationships with personified objects that are interpreted by mothers as vertical. The higher proportion of children with personified objects than of children with invisible friends who provided their companions with caregiving and guidance suggests otherwise. The notion that children with imaginary companions may be creating relationships for themselves is supported by mothers’ relationship-related explanations of invisible imaginary companions. The explanation that received the highest proportion of endorsements from mothers of these children was the child’s need for a relationship. In addition, unlike mothers of children with personified objects, between one-third and onefifth of the mothers of children with invisible companions explained their children’s friends in part as a function of a lack of playmates, a change in the family, and the child’s birth order. The last of these explanations is especially interesting given the high proportion of children with invisible companions who were firstborn or only children. Mothers of children with personified objects did not commonly refer to relationships to explain their children’s imaginary companions, but this finding may be due to mothers’ tendency to explain the existence of personified objects as the result of acquisition of the object. Such an explanation does not account for why the relationship was created and may have minimized the range of alternative justifications mothers provided for these companions. Mothers of children with invisible companions, in contrast, had no singular event that accounted for their children’s imaginary friends, and thus they may have felt prompted to provide more complex interpretations. Finally, the high incidence of personified objects, and of less animated transitional objects, in the preschool population makes their creation less anomalous and therefore less evoking of explanations than the creation of the rarer invisible friends. Suspecting that mothers might provide more extensive explanations for invisible companions than for personified objects, especially if mothers were ambivalent about the companions’ existence, we asked them to discuss their reactions to their children’s imaginary companions. Only one parent whose child had an invisible companion reported actively trying to discourage the phenomenon. A higher proportion of mothers of children with personified objects (67%) than of mothers of children with invisible companions (50%) reported encouraging the existence of their child’s imaginary companion rather than maintaining neutrality, although this difference was not statistically significant. At first glance, our call for further discrimination of invisible friends and personified objects appears to contradict Taylor and Carlson’s (1997) findings regarding the similar performance of children with different types of imaginary companions on theory-of-mind tasks. The findings presented in this study, however, although emphasizing the differences in the manifestations of invisible friends and personified objects and the different ways in which children relate to them, do not address the cognitive underpinnings of the creation of imaginary companions considered by Taylor and Carlson. Taken together, the present research and Taylor and Carlson’s work suggest that although the representational skills necessary for forming either type of imaginary companion are the same, the ways in which these skills are applied and the resulting imaginary companions may be different. Grouping children with imaginary companions
Imaginary companions of preschool children 107 of either type may therefore be a sensible approach if one’s goal is to differentiate them in terms of cognitive skills from children without imaginary companions. Such grouping may not be prudent, however, if one’s aim is to explain the developmental significance of imaginary companions in the social domain. Distinguishing Children with Invisible Friends, Children with Personified Objects, and Children with No Imaginary Companion The children with invisible friends and personified objects in this sample did not differ from the children without imaginary companions in having friends and positive reactions to preschool. The fact that the only social environment factors that differentiated children with and without imaginary companions were birth order and number of siblings confirms the expectation that social experiences within the family have a greater influence on imaginary companion creation than do social experiences outside the family. When considered in conjunction with mothers’ explanations, these results imply that for children with personified objects, the notion that an imaginary friend functions to compensate for a lack of relationships seems particularly unlikely. Even though twothirds of these children were only children or firstborn in their families, their mothers did not associate the creation of their companions with the absence of playmates (although the potential limitations of mothers’ explanations have been discussed). The compensatory hypothesis seems more plausible for children with invisible friends, because most of these children were first-born or only children and one-third of their mothers explained the creation of their imaginary companions in terms of relationships.
FUTURE DIRECTIONS Future research on the relationships that children form with imaginary companions should be conducted similarly to the way in which research on children’s real relationships is conducted, that is, through interviews and observation. Context, in particular, might be an important influence on children’s relationships with invisible friends and personified objects, just as it is in real relationships. Qualitative research could address the fact that little information is available concerning the times, places, and ways in which imaginary companions appear in children’s daily lives. Such accounts could distinguish further between invisible friends and personified objects while simultaneously offering suggestions as to why children create imaginary companions. As for the correlates of the formation of imaginary companions, the wellestablished link between birth order or number of siblings and the presence of imaginary companions implies that more attention should be directed toward the social provisions of sibling relationships as predictors of the formation of imaginary companions. In addition, the imaginary companion types may have differentiated precursors that would go undetected unless children with different types of imaginary companions are examined separately.
Annual progress in child psychiatry and child development 2000–2001
108
CONCLUSION Investigation into the qualities of the relationships that children have with imaginary others has the potential to explain further the developmental significance of pretend friends. Although the exact nature of the differentiated functions of invisible companions and personified objects cannot be ascertained from the present research, the results indicate that, at a minimum, invisible friends and personified objects should be investigated separately. Detection of subtle differences in the functions of these fantastical beings will only be possible if the apparent differences in their characteristics and in the ways in which children interact with them are acknowledged. Furthermore, distinctions between children with and without imaginary companions appear to be a function, at least in part, of a child’s familial context.
ACKNOWLEDGMENTS We wish to thank Kathryn Tout and Megan Gunnar for their help with this study; several undergraduate research assistants, especially Jennifer McGinley and Raechel DesCombaz, for data collection and coding; the mothers who participated in this study; and the teachers and staff at the preschools.
REFERENCES Ames, L., & Learned, J. (1946). Imaginary companions and related phenomena. Journal of Genetic Psychology, 69, 147–167. Bender, L., & Vogel, F. (1941). Imaginary companions of children. American Journal of Orthopsychiatry, 11, 56–65. Benson, R., & Pryor, D. (1973). “When friends fall out”: Developmental interference with the function of some imaginary companions. Journal of the American Psychoanalytic Association, 21, 457–473. Fulmer, A. (1995). Predictors of imaginary companions. Unpublished master’s thesis, University of California, Davis. Harter, S., & Chao, C. (1992). The role of competence in children’s creation of imaginary friends. Merrill-Palmer Quarterly, 38, 350–363. Harvey, N. (1918). Imaginary playmates and other mental phenomena of children. Ypsilanti, MI: State Normal College. Hurlock, E., & Burstein. M. (1932). The imaginary playmate: A questionnaire study. Journal of Genetic Psychology, 41, 380–392. Jersild, A., Markey, F., & Jersild, C. (1933). Children’s fears, dreams, wishes, daydreams, likes, dislikes, pleasant and unpleasant memories. New York: Teacher’s College, Columbia University. Manosevitz, M., Fling, S., & Prentice, N. (1977). Imaginary companions in young
Imaginary companions of preschool children 109 children: Relationships with intelligence, creativity and waiting ability. Journal of Child Psychology and Psychiatry, 18, 73–78. Manosevitz, M., Prentice, N., & Wilson, F. (1973). Individual and family correlates of imaginary companions in preschool children. Developmental Psychology, 8, 72–79. Masih, V. (1978). Imaginary play companions of children. In R.Weizman, R.Brown, P.Levinson, & P.Taylor (Eds.), Piagetian theory and the helping professions (pp. 136– 144). Los Angeles, CA: University of Southern California Press. Mauro, J. (1991). The friend that only I can see: A longitudinal investigation of children’s imaginary companions. Unpublished doctoral dissertation, University of Oregon. Meyer, J., & Tuber, S. (1989). Intrapsychic and behavioral correlates of the phenomenon of imaginary companions in young children. Psychoanalytic Psychology, 6, 151–168. Milne, A.A. (1924). When we were very young. New York: Dutton. Partington, J., & Grant, C. (1984). Imaginary playmates and other useful fantasies. In P.Smith (Ed.), Play in animals and humans (pp. 217–240). New York: Basil Blackwell. Schaefer, C. (1969). Imaginary companions and creative adolescents. Developmental Psychology, 1, 747–749. Singer, D., & Singer, J. (1990). The house of make-believe. Cambridge, MA: Harvard University Press. Singer, J., & Singer, D. (1981). Television, imagination, and aggression: A study of preschoolers. Hillsdale, NJ: Erlbaum. Singer, J., & Streiner, B. (1966). Imaginative content in the dreams and fantasy play of blind and sighted children. Perceptual and Motor Skills, 22, 475–482. Somers, J., & Yawkey, T. (1984). Imaginary play companions: Contributions of creative and intellectual abilities of young children. Journal of Creative Behavior, 18, 77–89. Spelling, O. (1954). An imaginary companion, representing a prestage of the superego. Psychoanalytic Study of the Child, 9, 252–258. Svendsen, M. (1934). Children’s imaginary companions. Archives of Neurology and Psychiatry, 32, 985–999. Taylor, M. (1999). Imaginary companions and the children who create them. New York: Oxford University Press. Taylor, M., & Carlson, S. (1997). The relation between individual differences in fantasy and theory of mind. Child Development, 68, 436–455. Taylor, M., Cartwright, B., & Carlson, S. (1993). A developmental investigation of children’s imaginary companions. Developmental Psychology, 29, 276–285. Vostrovsky, C. (1895). A study of imaginary companions. Education, 15, 393–398. Winnicott, D. (1953). Transitional objects and transitional phenomena. International Journal of Psychoanalysis, 34, 89–97.
Part II PARENTING
PART II: PARENTING
7 Contemporary Research on Parenting: The Case for Nature and Nurture W.Andrew Collins, Eleanor E.Maccoby, Laurence Steinberg, E.Mavis Hetherington, and Marc H.Bornstein
Current findings on parental influences provide more sophisticated and less deterministic explanations than did earlier theory and research on parenting. Contemporary research approaches include: (a) behavior-genetic designs, augmented with direct measures of potential environmental influences; (b) studies distinguishing among children with different genetically influenced predispositions in terms of their responses to different environmental conditions; (c) experimental and quasi-experimental studies of change in children’s behavior as a result of their exposure to parents’ behavior, after controlling for children’s initial characteristics; and (d) research on interactions between parenting and nonfamilial environmental influences and contexts, illustrating contemporary concern with influences beyond the parent-child dyad. These approaches indicate that parental influences on child development are neither as unambiguous as earlier researchers suggested nor as insubstantial as current critics claim.
INTRODUCTION The heredity and environment of an organism can be completely separated only in analytic thinking, for in actual nature such separation would lead to instant death of the organism, even though the philosopher making the analysis might himself survive (Gesell & Thompson, 1934, p. 293). Research on parenting has been the centerpiece of long-standing efforts in psychology to understand socialization processes. As the field moves into its second century, however, this focus on parental influence faces several highprofile challenges. One challenge comes from the charge that there is little compelling evidence of parents’ Jerome Kagan served as action editor for this article.
Annual progress in child psychiatry and child development 2000–2001
114
influence on behavior and personality in adolescence and adulthood (Harris, 1995, 1998; Rowe, 1994). Another is the allegation that socialization researchers have neglected significant forces other than parenting—forces that may contribute more extensively than parenting to individual differences in adult behavior. The most commonly cited sources of alternative influences are heredity (Harris, 1995, 1998; Rowe, 1994) and peers (Harris, 1995, 1998), although some writers emphasize the relatively greater importance of concurrent environmental forces more generally (e.g., Lewis, 1997). These criticisms of socialization research generally invoke studies of parenting published before the early 1980s. Neither the assumptions nor the research paradigms that dominated the field as recently as a decade ago, however, represent research on parenting today. Contemporary students of socialization largely agree that early researchers often overstated conclusions from correlational findings; relied excessively on singular, deterministic views of parental influence; and failed to attend to the potentially confounding effects of heredity. Contemporary researchers have taken steps to remedy many of those shortcomings. Unfortunately, the weaknesses of old studies still permeate presentations of socialization research in introductory textbooks and the mass media, partly because they appeal to preferences for simple generalizations instead of the conditional effects that capture the reality of socialization. Leading-edge approaches to social development and personality no longer rely exclusively on correlational designs, overly simple laboratory analogs, or additive models for assigning variance to one source or another. Contemporary studies, including research on parenting, turn on complex statistical methods and research designs that capture realworld complexity without sacrificing the rigor necessary to infer causal relations. Moreover, conceptual models increasingly encompass multiple sources of influence. Researchers draw on emerging knowledge in behavior genetics, neuroendocrine studies, studies of animal behavior, and intervention and prevention science to recognize the complex interplay between inherited and experiential components of individual development. The result is both a more complete and a more differentiated picture of parenting and its likely effects (for comprehensive reviews of contemporary socialization research, see Bornstein, 1995b; Eisenberg & Damon, 1998). One goal of this article is to outline key features of contemporary approaches to studies of parental socialization. We also show how current researchers have, for some time, been identifying and responding to the very challenges pointed to by recent critics. We pay particular attention to research designs that estimate inherited and other dispositional factors, as well as experiential ones, in estimating influence. We describe several lines of evidence that address issues of causality regarding the scope and nature of parental influences. Finally, we propose that responsible conclusions about the significance of parenting can be based on only the emerging body of research findings that incorporate both individual and social factors and their interrelations.
CONTEMPORARY APPROACHES TO PARENTING RESEARCH Research during the past two decades has undermined the once tacit assumption that environment should be the sole starting point in explaining individual differences in
Contemporary research on parenting 115 development. The relevant evidence comes from comparisons of the degree of similarity between individuals who vary in degree of genetic relatedness (e.g., identical vs. fraternal twins). Typical results imply that heredity accounts for a substantial proportion of this similarity, even though a recent meta-analysis (McCartney, Harris, & Bernieri, 1990) concluded that heredity rarely accounts for as much as 50% of the variation among individuals in a particular population, perhaps even less when personality characteristics are the focus. Although these findings also imply that environment contributes substantially to individual differences, behavior-genetics researchers typically infer environmental effects from the residual after estimates of genetic contributions are computed. The sources of the apparent environmental influences are not specified. Efforts to understand the role of parents in socialization are constrained severely by the traditional analytic model on which the most cited behaviorgenetic findings are based. This “additive” model regards hereditary and environmental components as independent and separable and holds that these two components together account for 100% of the variance in a characteristic (Plomin, 1990). Consequently, most behavior-genetic research has allowed for only main effects of genes and environment, ignoring the possibility that genes may function differently in different environments. A primary problem in disentangling heredity and measures of environmental influences, however, is that genetic and environmental factors are correlated (Plomin, 1990). Researchers consistently find that parenting of identical twins is more similar than parenting of fraternal twins and that two biological siblings typically experience more similar parenting than do two adopted children (Dunn & Plomin, 1986; Plomin, DeFries, & Fulker, 1988; Reiss, Niederhiser, Hetherington, & Plomin, in press; Rowe, 1983). Parents’ genotypes, as well as children’s genotypes, contribute to these contrasting patterns. That individuals who are more closely related genetically also have more similar shared parental environments means that observed associations between parenting and measures of child characteristics cannot be assumed to be either entirely genetic or entirely environmental in origin. As Rose (1995) stated it, the central question in development is “how genetic effects are modulated across lifespans of environmental interactions” (p. 627). A related problem further limits the usefulness of traditional behavior-ge-netic approaches to research on parenting. Estimating the effects of heredity versus environment ignores the potential for malleability, even in characteristics heavily influenced by heredity. When environmental conditions change substantially over time, mean levels of a characteristic also may change, although heritability coefficients (which are based on correlations) may or may not change (Plomin & Rutter, 1998). The problem comes from the failure to recognize that means and correlations can vary independently. Thus, although intelligence has been shown to have a high heritability coefficient, individuals’ cognitive abilities can improve or decline as a function of experience (for an explanation of this point, see Weinberg, 1989). Migration studies often reveal similar paradoxes. For example, height is highly heritable, with heritability coefficients in the .90s, showing that within a given population, the variation in children’s heights is closely linked with the variations in their parents’ heights. By inference, very little variance remains to be attributed to environmental factors. At the same time, grandparents born in Japan are, on the average, considerably shorter than their grandchildren born and reared in the United States
Annual progress in child psychiatry and child development 2000–2001
116
(Angoff, 1988). In the same way, genetic factors that are highly important in a behavior do not show up in a study of the heritability of that behavior because this genetic factor is uniform for all members of a population. Thus, analyzing the variation of a factor within a population does not provide exhaustive information concerning either the genetic or the environmental contributions to the factor. Large-scale societal factors, such as ethnicity or poverty, can influence group means in parenting behavior—and in the effects of parenting behaviors—in ways that are not revealed by studies of within-group variability. In addition, highly heritable traits also can be highly malleable. Like traditional correlational research on parenting, therefore, commonly used behavior-genetic methods have provided an incomplete analysis of differences among individuals. To acknowledge the importance of the interplay of heredity and environment, four lines of contemporary research on parenting have emerged. One line of research adopts the additive model of behavior-genetics research but augments it with direct measures of potential environmental influences in an effort to document environmental effects more precisely (Plomin et al., 1988; Reiss et al., in press). A second line of research addresses the insensitivity of additive models to Gene×Environment effects (Plomin & Rutter, 1998; Rutter et al., 1997) by distinguishing among children with different genetic predispositions on a characteristic to see whether they respond differently to different environmental conditions. The distinctions among genetically different groups often rely on measures of temperament or the parent’s carrying a known genetic risk factor. A third line of research examines the effect of parental practices after controlling for any initial dispositional characteristics of children. This kind of research is intended to permit inferences about the direction of effects when parent and child characteristics are initially correlated. Evidence on this point comes from three types of research designs: (a) longitudinal stud-ies in which child characteristics at Time 1 are controlled statistically; (b) experiments in which nonhuman animals are exposed to selected rearing environments; and (c) intervention studies either in which “experiments of nature” have resulted in marked changes in parenting experiences or in which families are randomly assigned to different treatment programs designed to improve parenting with resulting changes in child behavior. A fourth line of contemporary studies addresses the possibility that extrafamilial environmental conditions with which parenting is correlated contribute to individual differences in development and behavior. Augmented Behavior-Genetic Designs Traditional behavior-genetic designs give primacy to the effects of heredity, relying on a series of computations to reveal which portions of the variance should be labeled as contributions of the shared environment or assigned to nonshared, “other,” or “unknown” sources. Although evidence of shared family influences and experiences has appeared for some characteristics such as health habits, alcohol patterns, smoking patterns (McGue, 1994), depression in later life (Gatz, Pedersen, Plomin, Nesselroade, & McLearn, 1992), delinquency as reported by siblings (Rowe, Chassin, Presson, Edwards, & Sherman, 1992), and autonomy and sociability (Reiss et al., in press), the most frequent conclusion has been that shared environments play a small, inconsequential role in children’s development.
Contemporary research on parenting 117 Many scholars, however, have challenged this inference. One criticism is that the assumptions, methods, and truncated samples used in behavior-genetic studies maximize the effects of heredity and features of the environment that are different for different children and minimize the effects of shared family environments (Goodman, 1991; Hoffman, 1991; Patterson, 1999; Rose, 1995; Stoolmiller, 1999). For example, Stoolmiller (1999) noted that recent adoption studies have been impaired by pronounced range restrictions (about 67%) in the family environments sampled. Stoolmiller argued that the estimated contribution of shared environment likely would be as much as 50% higher if appropriate corrections for range restriction were applied to data from such studies. A second criticism is that estimates of the relative contributions of environment and heredity vary greatly depending on the source of data (Turkheimer & Waldron, in press). Twin studies typically yield higher heritability estimates for a trait than adoption studies do (Wachs & Plomin, 1991). Moreover, in both types of studies, heritability estimates vary considerably depending on the measures used to assess similarity between children or between parents and children. The largest effect sizes for environmental influences on social development are found with the relatively rarely used method of direct behavioral observations, whereas the smallest effect sizes for environmental influences are found with parental reports, which are the most commonly used measure in behavior-genetic studies of behavioral outcomes (Emde et al., 1992; GhodsionCarpey & Baker, 1997; Miles & Carey, 1997; Rutter et al., 1997; Wachs, 1992). The sizable variability in estimates of genetic and environmental contributions depending on the paradigms and measures used means that no firm conclusions can be drawn about the relative strength of these influences on development. Traditional twin and adoption studies have been criticized on the grounds that they estimate environmental effects only as a residual: the effects remaining after genetic effects have been estimated and subtracted from 100%. Efforts to rectify this problem by measuring environment directly, however, have failed to clarify the contributions of environment relative to heredity. Most such efforts were stimulated by Plomin and Daniels’ (1987) proposal that the environmental variance in behavior-genetic studies emanates largely from experiences that differ for children in the same family. By measuring such differences, researchers hoped to better understand the portion of the variance in behavior-genetic studies not attributed specifically to genetic relatedness. Behavior-genetic analyses, however, can establish that nonshared environment contributes to individual differences in a domain but cannot document the connections between objectively measured nonshared environmental events and development (Turkheimer & Waldron, in press). Most studies with direct measures of the environment and the development of multiple siblings within a family, moreover, have not used designs that permit heritability estimates (e.g., Brody & Stoneman, 1994; Tejerina-Allen, Wagner, & Cohen, 1994). Thus, researchers’ attempts to work within the traditional additive model, while augmenting it with direct measures of environment, have yielded findings that are conditional on a series of methodological problems in assessing the relevant environmental factors and in the inherent limitations of the additive model for identifying Gene×Environment interactions. The remainder of this chapter is devoted to recent
Annual progress in child psychiatry and child development 2000–2001
118
investigations of how processes of influence operate and interact. The Search for Gene×Environment Effects Traditional behavior-genetic models do not afford comparisons of the effects of differing environments on individuals who vary on genetically influenced characteristics. For example, in twin and adoption studies, degree of biological relatedness between individuals is the primary focus, not specific markers of genetically linked characteristics in the two individuals, and variations in environments are rarely assessed. The most likely possibility is that the forced estimates of main effects for genetic relatedness and environment in the additive model mask virtually ubiquitous correlations and statistical interactions between the two in existing research. Such interactions are notably diffi-cult to detect because of low statistical power in most relevant studies (McCall, 1991; McClelland & Judd, 1993; Wahlsten, 1990). Although some writers (e.g., Harris, 1998) have elected to subsume evidence of Gene×Environment correlations and interactions under genetic contributions to behavioral development, responsible scholarship requires closer attention to emerging evidence that these effects involve direct parental influences as well (O’Connor, Deater-Deckard, Fulker, Rutter, & Plomin, 1998; Plomin & Rutter, 1998). The search for Gene×Environment effects often takes the form of using measures of temperament for the purpose of distinguishing among children with different genetic predispositions to see whether they respond differently to given environmental conditions (Bornstein, 1995b; Plomin & Rutter, 1998; Rutter et al., 1997). Studies that pool parenting effects across children with very different temperaments inevitably obscure actual parental effects. Even when parenting effects are apparent, it is not reasonable to expect that a given style or quality of parenting would have the same effect on every child. Moreover, different parental strategies or degrees of parental effort may be required to bring about the same outcome in different children. Two types of recent studies attempt to disentangle individual children’s heredity and the nature of their rearing experiences: (a) studies of the effect of rearing experiences on the behavior of children who differ on measures of temperament; and (b) studies comparing the effect of high- versus low-risk environments on children of differing vulnerability. Temperament and Parenting. Temperamental characteristics, defined as “constitutionally based individual differences in reactivity and self-regulation” (Rothbart & Ahadi, 1994, p. 55), are thought to emerge early, to show some stability over time, but to be modifiable by experience. In general, statistical associations between early temperamental characteristics and later adjustment are modest (see Rothbart & Bates, 1998, for a review), suggesting that these associations also may be moderated by environmental factors. A difficult temperament, characterized by intense negative affect and repeated demands for attention, is associated with both later externalizing and internalizing disorders (Bates & Bayles, 1988; Bates, Bayles, Bennett, Ridge, & Brown, 1991). Early resistance to control, impulsivity, irritability, and distractibility predicts later externalizing and social alienation (Caspi, Henry, McGee, Moffitt, & Silva, 1995; Hagekull, 1989, 1994), whereas early shy, inhibited, or distress-prone behaviors predict later anxiety disorders, harm avoidance, and low aggression and social potency (Caspi &
Contemporary research on parenting 119 Silva, 1995). Correlations between temperamental characteristics and parental behavior reflect bidirectional interactive processes, as well as genetic linkages between parent and child characteristics. Temperamental characteristics may set in motion a chain of reactions from others that put children at risk or protect them from developing behavior and psychological problems (Caspi & Elder, 1988; Hetherington, 1989, 1991; Quinton, Pickles, Maughan, & Rutter, 1993; Rutter, 1990; Rutter & Quinton, 1984; Werner, 1990). Difficultness, irritability, and distress proneness in infants evoke hostility, criticism, a tendency to ignore the child, avoidance, coercive discipline, and a lack of playfulness in mothers (Lee & Bates, 1985; Rutter & Quinton, 1984; Van den Boom, 1989). These reactions, in turn, are associated with avoidant (Grossman, Grossman, Spangler, Suess, & Unzner, 1985; Van den Boom, 1989) or insecure-ambivalent attachment (Goldsmith & Alansky, 1987; Miyake, Chen, & Campos, 1985). Bates, Pettit, and Dodge (1995), in a longitudinal study, found that infants’ characteristics (e.g., hyperreactivity, impulsivity, and difficult temperament) significantly predicted externalizing problems 10 years later. Although this finding at first seems to support the lasting effects of physiologically based characteristics, Bates et al. (1995) also showed that predictive power increased when they added information about parenting to the equation. Infants’ early characteristics elicited harsh parenting at age 4, which in turn predicted externalizing problems when the children were young adolescents, over and above the prediction from infant temperament. Similarly, this and other findings imply that even though parental behavior is influenced by child behavior, parents’ actions contribute distinctively to the child’s later behavior. For example, in a longitudinal adoption design, O’Connor et al. (1998) confirmed that children at genetic risk for antisocial behavior elicited more negative parenting from adoptive parents than did children not at risk. They also found, however, that “most of the association between negative parenting and children’s externalizing behavior was not explicable on the basis of an evocative gene-environment correlation and that an additional environmentally mediated parental effect on children’s behavior was plausible” (p. 970). Bidirectional and interactive effects of this kind now appear to carry significant implications for distinctive effects of parenting variations on children who differ in temperamental characteristics. In longitudinal work on the socialization of “conscience,” Kochanska (1995, 1997) found that maternal use of gentle childrearing techniques that deemphasized power assertion was more effective with temperamentally fearful children than with bolder, more exploratory children in promoting the development of conscience. With bolder children, maternal responsiveness and a close emotional bond with the child were more important in fostering conscience. Similarly, the quality of parenting to some extent moderates associations between early temperamental characteristics of difficultness, impulsivity, and unmanageability and later externalizing disorders (Bates, Pettit, Dodge, & Ridge, 1998; Rothbart & Bates, 1998). Firm, restrictive parental control has been linked to lower levels of later externalizing in early difficult, unmanageable children (Bates et al., 1998). Although only a few studies have examined the moderating effects of parenting on the links between temperamental predispositions and later adjustment, and although not all of these studies have had positive results (Rothbart & Bates, 1998), the evidence nevertheless suggests that parenting moderates these
Annual progress in child psychiatry and child development 2000–2001
120
associations. Studies of Risk and Resiliency. Parallels to these differential relations between parenting and child behavior can be found in studies of risk and resiliency. Children who showed early developmental problems because of risk factors such as perinatal damage (Werner & Smith, 1992) improved in adjustment under authoritative parenting. Parenting, moreover, appears to play a mediating role between parental psychopathology and child symptoms of disorder (R.Conger, Ge, Elder, Lorenz, & Simons, 1994; Ge, Conger, Lorenz, Shanahan, & Elder, 1995; Ge, Lorenz, Conger, Elder, & Simons, 1994). For example, Downey and Walker (1992) demonstrated that children with a psychiatrically ill parent who were not exposed to parental maltreatment, in contrast to those who were, showed very low levels of both externalizing and internalizing. That different outcomes for children are associated with differential parental responses to the same risk factor implies parental influence, although Downey and Walker cannot rule out evocative behavior on the part of the child. A Finnish adoption study (Tienari et al., 1994) further illustrates how a genetic predisposition can either manifest itself or not, depending on whether certain triggering environmental conditions are present. Adoptees who had a schizophrenic biological parent were more likely to develop a range of psychiatric disorders (including schizophrenia) than were adoptees not at genetic risk, but only if they were adopted into dysfunctional families (see also Cadoret, 1985). Similar findings have been reported from studies of adopted children whose biological parents had a history of criminality (Bohman, 1996). If adopted into well-functioning homes, 12% of these children displayed petty criminality in adulthood. However, if adopted into families carrying environmental risk, their rate of petty criminality in adulthood rose to 40%. These findings suggest that well-functioning parents can buffer children at genetic risk and circumvent the processes that might ordinarily lead from genotype to phenotype. The more general point is that genetic vulnerabilities (or strengths) may not be manifested except in the presence of a pertinent environmental trigger such as parenting. Studies of Parental Influence, Controlling for Initial Child Characteristics A third line of research attempts to provide a basis for examining instances in which parental behavior may exert a causal influence in changing children’s behavior. Studies of this type subsume several research strategies. One strategy is longitudinal research in which children’s initial characteristics can be observed to change over time in relation to specific parenting experiences. Even more compelling evidence for determining the causal status of parenting, however, involves experimental manipulations. In some recent experiments, young nonhuman animals were exposed to measurably different rearing con-ditions. Some experiments of nature with humans also have provided evidence of this kind. The most compelling evidence, however, comes from interventions in which parents are assigned randomly to behavior-change treatment groups, with resulting changes in the behavior of both the parents and their otherwise untreated children. Random assignment is the means for ensuring that treatment groups are not initially different. Longitudinal Studies of Parenting and Child Development. The most widely used
Contemporary research on parenting 121 strategy in contemporary studies of socialization uses short-term longitudinal designs to better distinguish parenting effects from the characteristics of the child (e.g., Ge et al., 1996; Steinberg, Lamborn, Darling, Mounts, & Dornbusch, 1994). In these studies, aspects of child functioning and development are measured at more than one point in time. Statistical procedures, such as the analysis of covariance or multiple regression, are then used to estimate the relation between parenting at one point in time and child outcomes at some subsequent point, after taking into account characteristics of the child at the time that parenting was assessed. Studies showing that the over-time effect of parenting on child development holds even after controlling for earlier child characteristics are important for several reasons. First, in the absence of a randomized experimental design, this strategy provides indirect evidence that parenting conceivably affects—rather than simply accompanying or following from—child adjustment. Such indirect evidence is important because one cannot randomly assign children to different home environments. These analyses do not rule out the possibility that different children elicit different parental responses, but they do provide evidence that the correlation between child adjustment and parenting is not due solely to the effect of children on parenting behavior. Significant longitudinal relations between parenting and child adjustment after taking into account their concurrent relation also help to rule out a number of third-variable explanations, including the possibility that the observed association is due to factors that parents and their children share, such as genes or socioeconomic status. To be a viable explanation for the observed association, a third variable would have to be correlated with the measures of child adjustment at the time of the longitudinal follow-up but not correlated with the same measures taken earlier. Any genetically mediated link between parenting and child adjustment, for example, would be taken into account by controlling for the concurrent relation between parenting and child adjustment before examining their relation over time. Rearing Experiments with Animals. Recent work with nonhuman animals points clearly to the fact that experience—that is, encountering or engaging with the environment—influences brain development in young organisms and that these changes in the brain are associated with changes in behavior (Greenough & Black, 1992). Although some of the relevant environmental events must occur during a sensitive period to affect development (Bornstein, 1989), the mammalian brain generally remains malleable by environmental inputs well into adulthood (Huttenlocher, 1994; Nelson, in press). Environmental events that have to do with the amount or kind of “parenting” that a young organism receives are essential for survival in all mammalian species. The presence and activities of the infant stimulate a set of maternal behaviors needed by the infant (including but not confined to feeding), and these reciprocal maternal behaviors serve to facilitate the infant’s adaptation and development (e.g., Stern, 1985). Studies of higher mammals confirm that, as these interactions continue to occur, an intense emotional bond is formed such that separation of the pair produces distress and behavioral disruption in each member of the pair. Studies in which young animals have been deprived of “mothering” have shown clearly that such deprivation not only disrupts the ongoing behavior of the young animal at the time of deprivation but also leads to dysfunctional outcomes for the offspring in the long term.
Annual progress in child psychiatry and child development 2000–2001
122
Current animal work is addressing implications of naturally occurring variation, within the “normal” range, in maternal behavior. Meaney and Plotsky and their colleagues (Caldji et al., 1998; Liu et al., 1997) have studied styles of mothering in rats, relating variations in these styles to behavioral outcomes in their offspring. Maternal animals differ considerably in the frequency with which they lick and groom their newborn pups and in whether they arch their backs to facilitate nursing or lie passively on top of or next to the pups. Individual differences in these mothering styles have been shown to be quite stable. In adulthood, moreover, the offspring of mothers who had done more licking and grooming and had nursed with arched backs (high LG-ABN mothers, whom we can call nurturant) were less timid in leaving their home cages to obtain food or explore a novel environment than were the offspring of low LG-ABN mothers. These outcomes are correlated with neuroendocrine processes. As adults, rats who had experienced high levels of maternal licking and grooming as newborns showed reduced levels of adrenocorticotropic hormone and corticosterone in response to a stressful condition (close restraint). Furthermore, differences emerged in the densities of receptors for stress hormones in several loci in the brains of animals that had experienced the two different kinds of maternal styles in their first 10 days of life. Thus, early mothering styles apparently affected the neural circuitry that governs behavioral stress responses in the offspring as they grow into adulthood. To determine whether there is an independent effect of maternal styles per se on these outcomes, apart from any genetic mediation, researchers have crossfostered infants born to a low-nurturant mother to rearing by a highnurturant mother. Early findings (Anisman, Zaharia, Meaney, & Merali, 1998) show that these infants manifest the benefits of their early rearing in their modified adult stress reactions, by comparison with infants born to low-nurturant mothers and reared by them. Corroborating evidence comes from studies with nonhuman primates (Suomi, 1997). Suomi and colleagues initially observed naturally occurring individual differences in “emotional reactivity” among Rhesus monkeys. In early life, some animals are hesitant about exploring new environments and show extreme reactions to separation from their mothers, whereas others characteristically react more calmly. Individual animals’ reactivity patterns remain quite stable over many years. These patterns of behavior are accompanied by distinctive neuroendocrine patterns. The behavioral and physiological indicators that distinguish highly reactive animals from less reactive ones are especially apparent under environmentally stressful conditions (Suomi, 1997). When young Rhesus monkeys with clearly different reactivity patterns are crossfostered to mothers who are either reactive (easily distressed) or nonreactive (calm), their adult behavior is quite different from that shown by the biological offspring of each type of mother. Genetically reactive young animals that are reared by calm mothers for the first 6 months of their lives and then placed in large social groups made up of peers and nonrelated older adults develop normally and indeed rise to the top of their dominance hierarchy. Further, these cross-fostered animals are adept at avoiding stressful situations and at recruiting social support that enables them to cope with stress. By contrast, genetically reactive infants who are reared by reactive mothers typically are socially incompetent when placed in the larger living group at the age of 6 months and are particularly vulnerable to stress. In general, the introduction of stressful conditions seems
Contemporary research on parenting 123 to make the effects of early rearing experience especially perceptible (Suomi, 1997). Thus, variations in mothering style have a lasting effect on the reactivity of the young animals when they move into new social contexts. Moreover, the quality of early mothering now has been found to affect the way genetically at-risk females parent their own offspring. If crossfostered to low-reactive mothers, they are competent parents with their own offspring; if raised by high-reactive mothers, they manifest mothering deficits. Recent work (Suomi, in press) has shown that the genetic make-up of young monkeys influences how large an effect early rearing conditions will have. A gene has been identified for which one allele is associated with a highly reactive temperament and the other allele with a calmer temperament. Certain aspects of the neuroendocrine system (i.e., serotonergic functioning) are controlled by this gene. Maternal deprivation has a powerful effect on the genetically reactive monkeys, producing deficits in their neuroendocrine functioning and their behavioral and emotional reactions. For the animals not carrying the genetically risky allele, however, maternal deprivation has little effect. These recent studies trace some of the complex steps in the long pathway between genes and phenotypic behavior. The findings show that both genes and parenting affect brain processes and neuroendocrine systems. These studies point to a future in which researchers will be able to provide more detailed information about the interplay of heredity and parenting influences than traditional twin and adoption studies can yield. Experiments of Nature. No extensively controlled rearing experiments have been conducted with human children, but several natural experiments have yielded information that is strikingly parallel to the findings of the cross-fostering work. A recent example is found with the children who had lived in Romanian orphanages for some months or years in early childhood, during which time they were deprived of the opportunity to form a close bond with a single trusted adult caregiver. Some of these children have been adopted into middle-class homes in other cultures. The effects of the early deprivation appear to depend on its duration. Recent follow-up measures at age 6 in a group of Romanian orphans adopted by Canadian families show that children adopted during approximately the first half-year of life manifest no lasting effects of their early experience. But children adopted later have been found to have abnormally high levels of cortisol during the ordinary daily routine of their adoptive homes, indicating that the neuroendocrine system involved in stress regulation has not developed normally (Chisholm, 1998; Chisholm, Carter, Ames, & Morison, 1995; Gunnar, in press; see also Rutter & the ERA study team, 1998). An example of variations in parenting that are more within the normal range comes from France, where 20 children were located who had been abandoned in infancy by their low socioeconomic-status parents and adopted by uppermiddle-class parents (Schiff, Duyme, Dumaret, & Tomkiewitz, 1982). These children all had biological siblings or half-siblings who remained with the biological mother and were reared by her in impoverished circumstances. The researchers were unable to find any selective factors that might have made the abandoned children more genetically promising than the ones retained at home. When tested in middle childhood, however, the adopted children’s IQs averaged 14 points higher than those of their natural siblings. By contrast, children who remained with their biological mothers were four times more likely to exhibit failures in their school performance. These results are consistent with those of several other early
Annual progress in child psychiatry and child development 2000–2001
124
adoption studies (e.g., Scarr & Weinberg, 1976, 1978; Skodak & Skeels, 1949) showing that adoption into well-functioning middle-class homes can provide a “bonus” in cognitive functioning for the children involved. What aspects of living in more advantaged homes were responsible for these children’s cognitive and educational gains is not known. Was it the more stimulating, more cultured, more educated environments provided by the adoptive parents, or were there greater amounts of parent-child interaction or more secure attachments? We can only suspect that something about the way these adoptive parents dealt with the children contributed to the effect. Evidence from the Colorado Adoption Project provides some suggestive evidence for a bidirectional process. The Colorado project included data on rates of communicative development in groups of 12-months-olds either born or adopted into intact families (Hardy-Brown, 1983; Hardy-Brown & Plomin, 1985; HardyBrown, Plomin, & DeFries, 1981). Biological mothers’ verbal intelligence correlated with the language competencies of children they had not seen since birth. Reciprocally, however, adoptive mothers’ activities, like imitating their infants’ vocalizations and vocalizing responsively and contingently to infants’ vocalizations, also predicted child language competencies. Similarly, another comparison of children with their biological and their adoptive parents (Scarr & Weinberg, 1978) showed that correlations between the vocabulary scores of adoptive mothers and children were as high as those between the vocabulary scores of biological mothers and their children. Like other examples cited earlier, these findings clearly show the distinct contribution of parental behavior over and above the contribution of heredity. Interventions with Human Parents. Finally, interventions that seek to change the mean level of a behavioral or personality characteristic in children provide additional evidence of the efficacy of parenting. Efforts to manipulate parental behavior for the purpose of influencing child behavior are surprisingly rare. Laboratory analog studies (e.g., Kuczynski, 1984), although documenting shortterm effects of specific behaviors of parents, cannot establish that such behaviors significantly influence broadband outcomes for offspring. The primary source of relevant information for human children comes from evaluations of programs designed to remediate or prevent socialization problems. Such programs typically target the behavior of either children alone or both children and parents. Of particular relevance to socialization, however, are studies in which the behavior of parents, but not the children, is the target of the manipulation. If the manipulation produces desired changes in the parent’s behavior and if the degree of change, in turn, is associated with changes in the child’s behavior, the evidence for the causal influence of parents is compelling. Unfortunately, only a few such programs focus on improving parental behavior, and even fewer estimate the causal influences of changes in parental behavior on child outcomes (for reviews, see Cowan, Powell, & Cowan, 1998; McMahon & Wells, 1998). An exception is a recent prevention program intended to foster more effective parenting following divorce (Forgatch & DeGarmo, 1999). School-age sons of recently divorced single mothers often manifest increased academic, behavioral, social, and emotional problems relative to sons of nondivorced mothers, and the divorced mothers themselves commonly behave toward their sons in a more coercive and less positive manner than nondivorced mothers do (Chase-Lansdale, Cherlin, & Kiernan, 1995;
Contemporary research on parenting 125 Hetherington, 1993; Zill, Morrison, & Coiro, 1993). In most reports, however, the direction of causality is unclear. Forgatch and DeGarmo sought both to address the causality issue and to test a method for preventing these apparently negative sequelae of divorce. They designed group-intervention and individual follow-up procedures for 153 recently divorced mothers who met three criteria: They had been separated from their partners within the prior 3 to 24 months, they resided with a biological son in Grades 1 through 3, and they did not cohabit with a new partner. Initial observational, self-report, and teacher report measures of both mothers’ parenting and children’s behaviors were used to control for possible genetically influenced differences among parent-child pairs. Random assignment ensured that the treatment group was not systematically different from the control group of 85 mothers and sons who also met the screening criteria. No intervention was provided to the children. At the end of 12 months, treatmentgroup mothers generally showed less coercive behavior toward children and less decline in positive behavior than control-group mothers did (although both treatment and controlgroup mothers manifested at least temporary declines in positive behavior during the year following divorce). Moreover, the degree of change in the mothers’ behavior over the course of 12 months significantly predicted the degree of change in the children’s behaviors. Changes in parenting practices were associated significantly with changes in teacher-reported school adjustment and with changes in both child-reported and parentreported maladjustment. Estimated effect sizes for these correlated changes ranged from 0.032 to 0.144 (M.Forgatch, personal communication, November 1, 1999). These effect sizes are small to medium, according to Cohen’s (1988) criteria. Other intervention attempts with parents have yielded similarly impressive evidence. Cowan and Cowan (in press), in a randomized design, showed that parents’ participation in a 16-week series of discussion groups on effective parenting just prior to their children’s kindergarten entry resulted in better school adjustment and higher academic achievement for children in kindergarten and first grade, compared with children whose parents attended a series of discussion groups without the effective-parenting emphasis. The relative advantage for the children of intervention-group parents has persisted through age 10, a period of 6 years. With parents of infants, Van den Boom (1989, 1994) demonstrated that an intervention to train lower-class mothers to respond sensitively to their infants both modified the negative responses of mothers to infant irritability and reduced the extent of avoidant attachment in distress-prone infants. Similarly, Belsky, Goode, and Most (1980) found that interventions to increase mothers’ didactic interactions with infants during play resulted in significantly higher exploratory play among infants, compared with a no-treatment control group. In interventions to improve the behavioral-training skills of parents of noncompliant children, Forehand and colleagues demonstrated both improvements in parental behavior and behavioral changes in the children, as well as increased parental perceptions of improved child behavior and decreased parental depression (Forehand & King, 1977; Forehand, Wells, & Griest, 1980). Depending on the content of the maternal training, children have been shown to manifest differing patterns of competence. Riksen-Walraven (1978) showed that infants of mothers trained in responding demonstrated higher levels of exploratory competence, whereas infants of mothers trained on im-proving sensory stimulation habituated more efficiently. When interventions are effective, behavior change tends to be long-lasting
Annual progress in child psychiatry and child development 2000–2001
126
(Patterson, 1975). Findings from studies of parenting-focused interventions provide the strongest evidence available on the efficacy of parenting behavior in humans. Whether naturally occurring behaviors of the kind encouraged by these experimental programs account for behavioral development is more difficult to establish. Nevertheless, the increasing use of multimethod, multi-informant assessments and structural equation modeling is helping to overcome some of the shortcomings of traditional correlational studies of socialization and behavior-genetic studies using single informants (Rutter et al., 1997). These more methodologically rigorous studies (e.g., R.Conger & Elder, 1994; Forgatch, 1991; Kim, Hetherington, & Reiss, 1999) generally yield associations between parenting and child outcomes, with appropriate controls for Time 1 status on outcome measures, that meet Cohen’s (1988) criteria for small or medium effect sizes. Some studies (e.g., Kochanska, 1997) yield impressively large effect sizes. Even small effects of parenting, however, are likely to become large effects over time (Abelson, 1985). Parental behavior has been shown to be highly stable across time (Holden & Miller, 1999). Thus, specific parental influences, consistently experienced, likely accumulate to produce larger meaningful outcomes over the childhood and adolescent years. Studies of Links between Parenting and Other Influences Current investigations address a further challenge from recent critics of parenting research as well: The need to consider environmental influences other than parents in accounting for differences among children. Socialization research today is guided by an ecological perspective on human development (Bronfenbrenner, 1979; for recent reviews, see Bornstein 1995a, 1995b; Bronfenbrenner & Morris, 1998). Families are seen as important influences on children, the effect of which can be understood only in light of the simultaneous influence of social spheres such as peer groups and schools. These influences occur within broad contexts (e.g., neighborhood, cultural context, historical epoch) that add to, shape, and moderate the effect of the family. The ecological perspective not only emphasizes the potential significance of extrafamilial influences on the child’s development but also, more importantly, stresses the interactive and synergistic, rather than additive and competitive, nature of the links between the family and other influences. In this section we consider the implications of this view for parenting in relation to two extrafamilial influences on socialization: peers and macrocontexts of parent-child relations. Relations of Parental and Peer Influences. In an earlier era, socialization researchers cast families and peers as opposing forces vying for influence over the child’s behavior. In much the same way that recent developments in behav-ior genetics have challenged the wisdom of attempting to estimate how much variance in a trait is attributable to genes versus the environment, contemporary models of socialization no longer ask whether children are influenced more by parents or by peers. Today, socialization researchers develop and test models that examine how parents and peers exert conjoint influence on the developing child (e.g., Brown, Mounts, Lamborn, & Steinberg, 1993; Cairns & Cairns, 1994; Dishion, Patterson, Stoolmiller, & Skinner, 1991; Fuligni & Eccles, 1993; Mounts & Steinberg, 1995).
Contemporary research on parenting 127 This new direction rests on four findings that have emerged consistently from research on parent and peer influences. The first finding is that the observed similarity between adolescents and their friends across a wide array of variables, including school achievement (Epstein, 1983), aggression (Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988), internalized distress (Hogue & Steinberg, 1995), and drug use (Kandel, 1978), is due mostly to the tendency for individuals to select like-minded friends, as well as to the influence that friends have over each other (Berndt, 1999; Berndt, Hawkins, & Jiao, 1999). Children are not randomly assigned to peer groups. Although unambiguous estimates of the relative effect of selection and influence effects are not available, a child with antisocial inclinations may be far more likely to fall into a similarly inclined peer group than an antisocial peer group is to corrupt a wellbehaved youngster. Similarly, an academically oriented child may be more likely to select academically oriented friends than a child who is not interested in school is to develop a passion for achievement because his or her friends are so inclined. Equating peer influence with peer similarity overstates considerably the extent of peer influence, because the equation fails to take account of the selection effect (Bauman & Fisher, 1986). The second finding is that peer influence often operates with respect to everyday behaviors and transient attitudes, not enduring personality traits or values (Brown, 1990). Most studies examining individuals’ religiosity, educational plans, and occupational choices, for example, reveal that parental influence on adolescent personality development is deeper and more enduring than that of peers (Brown, 1990). To be sure, even transient peer influences over day-to-day behaviors can have enduring sequelae that are opposed to what parents might desire (e.g., peer influence to become sexually active can result in an unplanned pregnancy and foreshortened educational attainment; peer influence to engage in criminal activity can result in a jail sentence). However, because peer influence tends to be immediate, its content changes with shifts in friendships. Studies that track individuals through adolescence often reveal that young adults are more similar to their parents than they had appeared to be as teenagers (J.Conger, 1971). The third finding is evidence of the significance of parents and parent-child relationships in influencing which peers children select. Any psychological snapshot taken during adolescence, when peers are undeniably an important force in children’s lives, rightly should be viewed as the end of a long process of socialization that began early in childhood and most likely has its origins in the family. Parke and Bhavnagri (1989) indicated that parents influence children’s peer experiences in two general ways. During elementary school parents propel their children toward certain peers by managing their youngsters’ social activities (which has the effect of increasing contact with some peers and diminishing it with others); during both childhood and adolescence, parents actively steer children toward certain friends and away from others. In addition, throughout the child’s development parents indirectly influence the child’s attitudes, values, personality, and motives, which in turn affect the child’s interactions and affiliations with particular peers (Brown et al., 1993). For all of these reasons, parental and peer influence tend to be complementary, not antithetical (Brown, 1990). Finally, and perhaps most importantly, adolescents differ considerably in their susceptibility to peer influence, and one of the most important contributors to this differential susceptibility is the quality of the parent-child relationship. Adolescents
Annual progress in child psychiatry and child development 2000–2001
128
whose parents are authoritative (i.e., responsive and demanding) are less swayed by peer pressure to misbehave than are adolescents whose parents are permissive (Devereux, 1970) or authoritarian (Fuligni & Eccles, 1993). Indeed, adolescents from authoritative homes are more susceptible to prosocial peer pressure (e.g., pressure to do well in school) but less susceptible to antisocial peer pressure (e.g., pressure to use illicit drugs and alcohol; Mounts & Steinberg, 1995). In other words, the particular peers a youngster selects as friends and the extent to which he or she is susceptible to their influence are both affected by parenting. A compelling illustration of indirect effects of parents comes from research on the development of antisocial behavior and aggression (DeBaryshe, Patterson, & Capaldi, 1993; Dishion et al., 1991; Patterson, DeBaryshe, & Ramsey, 1989). Researchers consistently have confirmed that adolescents’ involvement in antisocial activity is influenced significantly by their relationships with antisocial peers but that the chain of events that leads some adolescents into antisocial peer groups begins at home during childhood. The links in this chain include exposure to harsh and coercive parenting, which contributes to the development of aggression and to academic difficulties in school; these problems, in late childhood, lead to the selection of antisocial peers. Even when selection effects are controlled, much of what appears to be peer influence is actually the end result of familial influence at an earlier point in the child’s development. Macrocontexts of Parenting. Parents also mediate the association between broader social, cultural, economic, and historical contexts and children’s behavior and personality. These broad contextual forces affect how parents behave and may accentuate or attenuate the effect of parental behavior on children’s development. R.Conger (e.g., R.Conger et al., 1994) and McLoyd (1990), for example, have demonstrated that many of the deleterious effects of poverty on children’s development are mediated through the effect of poverty on parenting; economic stress and disadvantage increase parental punitiveness, which in turn adversely affects the child. One implication of this for understanding the results of research on parenting is that estimates of the strength of parental influence are likely specific to particular communities in particular cultures at particular points in time. Many apparent “effects” of social class or economic disadvantage are mediated through the effect of these factors on parenting practices. An example comes from recent research on the effects of neighborhood contexts on children’s behavior and personality (Brooks-Gunn, Duncan, & Aber, 1997; BrooksGunn, Duncan, Klebanov, & Sealand, 1993; Chase-Lansdale & Gordon, 1996). Neighborhood characteristics have been shown both to influence parents’ behavior and to moderate the effect of parenting practices on the child’s development (Klebanov, BrooksGunn, & Duncan, 1994). The effect of neighborhoods on parental practices is evident in the finding that parents adjust their management strategies to suit the demands of the neighborhood context within which they live (Furstenberg, Eccles, Elder, Cook, & Sameroff, 1997). Parents who live in dangerous neighborhoods tend to be more controlling and restrictive, which protects the child’s physical well being but which also may have the unintended consequence of squelching the child’s sense of autonomy. With respect to moderating effects, Darling and Steinberg (1997) have shown that the links between parental involvement in school and children’s achievement vary as a function of the behavior of other parents in the neighborhood, with parental involvement having
Contemporary research on parenting 129 more potent effects within neighborhoods with high concentrations of involved parents. Similarly, the beneficial effects of authoritative parenting are accentuated when adolescents affiliate with peers who themselves have authoritative parents (Fletcher, Darling, Steinberg, & Dornbusch, 1995). The documented relations between parental and other influences are consistent with recent criticisms (e.g., Harris, 1995, 1998) that socialization researchers have overemphasized the role of parents and underemphasized the role of nonfamilial influences, most notably, the peer group. Studies of the broader context of parental socialization, however, neither support nor refute claims about the potency of parental influence. These studies amply illustrate that, far from a myopic focus on the influence of parents, contemporary researchers have for some time amassed evidence that socialization can be fully understood only by examining the role of parents in light of the influence of other settings in which children and families function.
CONCLUSIONS The lines of research just described imply a concept of parenting and parental influence that is more differentiated and complex than the dominant models of earlier eras. Whereas socialization researchers often depicted parents as “molding” children to function adequately in the society (Hartup, 1989; Maccoby, 1992), contemporary evidence clearly points toward multiple roles for parents that often do not imply the deterministic effect once attributed to them. Whereas researchers using behavior-genetic paradigms imply determinism by heredity and correspondingly little parental influence (e.g., Rowe, 1994), contemporary evidence confirms that the expression of heritable traits depends, often strongly, on experience, including specific parental behaviors, as well as predispositions and age-related factors in the child. Whereas both older traditions typically limited ideas about environmental effects to parents, contemporary researchers have shown the interrelated effects of parenting, nonfamilial influences, and the role of the broader context in which families live (e.g., Bronfenbrenner, 1979; Bronfenbrenner & Ceci, 1994; Brooks-Gunn et al., 1997; Darling & Steinberg, 1997; Wachs, 1999). This new generation of evidence on the role of parenting should add to the conviction, long held by many scholars, that broad, general main effects for either heredity or environment are unlikely in research on behavior and personality. Statistical interactions and moderator effects are the rule, not the exception. Information of this kind, unfortunately, fits poorly with the desire of the popular media for facile sound bites about parenting or the yearning of some writers of introductory textbooks for general, causal statements about behavioral development. Contrary to criticisms of socialization research, the difficulty today is not that the evidence is inadequate to show parenting effects but that the evidence has revealed a reality that is far more complex than critics expected or that writers can convey in most popular media outlets. For psychologists, the challenge is to make that reality a compelling foundation for the science and practice of the future and to find ways of disseminating this knowledge to a public eager to understand the forces that shape children’s development.
Annual progress in child psychiatry and child development 2000–2001
130
ACKNOWLEDGMENTS Preparation of this article was supported in part by the Rodney S.Wallace Professorship for the Advancement of Teaching and Learning, College of Education and Human Development, University of Minnesota. We thank the following for helpful comments on the manuscript: Marion S.Forgatch, Ben Greenberg, Megan R.Gunnar, Willard W.Hartup, Jerome Kagan, Gerald Patterson, Stephen Suomi, Deborah Vandell, Theodore Wachs, and Richard A. Weinberg.
REFERENCES Abelson, R. (1985). A variance explanation paradox: When a little is a lot. Psychological Bulletin, 97, 129–133. Angoff, W.H. (1988). The nature-nurture debate, aptitudes, and group differences. American Psychologist, 43, 713–720. Anisman, H., Zaharia, M.D., Meaney, M.J., & Merali, Z. (1998). Do early-life events permanently alter behavioral and hormonal responses to stressors? International Journal of Developmental Neuroscience, 16, 149–164. Bates, J., & Bayles, K. (1988). The role of attachment in the development of behavior problems . In J.Belsky & T.Nezworski (Eds.), Clinical implications of attachment (pp. 253–299). Hillsdale, NJ: Erlbaum. Bates, J., Bayles, K., Bennett, D.S., Ridge, B., & Brown, M.M. (1991). Origins of externalizing behavior problems at eight years of age. In E.J.Pepler & K.H. Rubin (Eds.), The development and treatment of childhood aggression (pp. 197–216). New York: Academic Press. Bates, J., Pettit, G., & Dodge, K. (1995). Family and child factors in stability and change in children’s aggressiveness in elementary school. In J.McCord (Ed.), Coercion and punishment in long-term perspectives (pp. 124–138). New York: Cambridge University Press. Bates, J., Pettit, G., Dodge, K., & Ridge, B. (1998). Interaction of temperamental resistance to control and restrictive parenting in the development of externalizing behavior. Developmental Psychology, 34, 982–995. Bauman, K., & Fisher, L. (1986). On the measurement of friend behavior in research on friend influence and selection: Findings from longitudinal studies of adolescent smoking and drinking. Journal of Youth and Adolescence, 15, 345–353. Belsky, J., Goode, M.K., & Most, R.K. (1980). Maternal stimulation and infant exploratory competence: Cross-sectional, correlational, and experimental analyses. Child Development, 51, 1168–1178. Berndt, T.J. (1999). Friends’ influence on children’s adjustment to school. In W.A. Collins & B.Laursen (Eds.), Relationships as developmental contexts: The Minnesota Symposia on Child Psychology (Vol. 30, pp. 85–108). Mahwah, NJ: Erlbaum. Berndt, T.J., Hawkins, J.A., & Jiao, Z. (1999). Influence of friends and friendship on
Contemporary research on parenting 131 adjustment to junior high school. Merrill-Palmer Quarterly, 45, 13–41. Bohman, M. (1996). Predispositions to criminality: Swedish adoption studies in retrospect. In G.R.Bock & J.A.Goode (Eds.), Genetics of criminal and antisocial behavior, Ciba Foundation Symposium 194 (pp. 99–114). Chichester, UK: Wiley. Bornstein, M.H. (1989). Sensitive periods in development: Structural characteristics and causal interpretations. Psychological Bulletin, 105, 179–197. Bornstein, M.H. (1995a). Form and function: Implications for studies of culture and human development. Culture and Psychology, 1, 123–137. Bornstein, M.H. (Ed.). (1995b). Handbook of parenting. Mahwah, NJ: Erlbaum. Brody, G., & Stoneman, Z. (1994). Sibling relations and their association with parental differential treatment. In E.M.Hetherington, D.Reiss, & R.Plomin (Eds.), Separate social worlds of siblings: The impact of nonshared environment on development (pp. 129–142). Hillsdale, NJ: Erlbaum. Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press. Bronfenbrenner, U., & Ceci, S.J. (1994). Nature-nurture reconceptualized in developmental perspective: A bioecological model. Psychological Review, 101, 568– 586. Bronfenbrenner, U., & Morris, P.A. (1998). The ecology of developmental processes. In W.Damon & R.M.Lerner (Eds.), Handbook of child psychology: Theoretical models of human development (5th ed., Vol. 1, pp. 993–1028). New York: Wiley. Brooks-Gunn, J., Duncan, G., & Aber, L. (Eds.). (1997). Neighborhood poverty: Context and consequences for children. New York: Russell Sage Foundation. Brooks-Gunn, J., Duncan, G., Klebanov, P., & Sealand, N. (1993). Do neighborhoods influence child and adolescent development? American Journal of Sociology, 99, 353– 395. Brown, B. (1990). Peer groups. In S.Feldman & G.Elliott (Eds.), At the threshold: The developing adolescent (pp. 171–196). Cambridge, MA: Harvard University Press. Brown, B., Mounts, N., Lamborn, S., & Steinberg, L. (1993). Parenting practices and peer group affiliation in adolescence. Child Development, 64, 467–482. Cadoret, R. (1985). Genes, environment and their interaction in the development of psychopathology. In T.Sakai & T.Tsuboi (Eds.), Genetic aspects of human development (pp. 165–175). Tokyo: Igaku-Shoin. Cairns, R., & Cairns, B. (1994). Lifelines and risks: Pathways of youth in our time. New York: Cambridge University Press. Cairns, R., Cairns, B., Neckerman, H., Gest, S., & Gariepy, J.L. (1988). Social networks and aggressive behavior: Peer support or peer rejection? Developmental Psychology, 24, 815–823. Caldji, C., Tannenbaum, B., Sharma, S., Francis, D., Plotsky, P.M., & Meaney, M.J. (1998). Maternal care during infancy regulates the development of neural systems mediating the expression of fearfulness in the rat. Proceedings of the National Academy of Science, 95, 5335–5340. Caspi, A., & Elder, G. (1988). Emergent family patterns: The intergenerational construction of problem behavior and relationships. International Journal of Behavioral Development, 5, 81–94.
Annual progress in child psychiatry and child development 2000–2001
132
Caspi, A., Henry, B., McGee, R.O., Moffitt, T.E., & Silva, P.A. (1995). Temperamental origins of child and adolescent behavior problems: From age 3 to age 15. Child Development, 66, 55–68. Caspi, A., & Silva, P. (1995). Temperamental qualities at age 3 predict personality traits in young adulthood: Longitudinal evidence from a birth cohort. Child Development, 66, 486–498. Chase-Lansdale, P.L., Cherlin, A., & Kiernan, K. (1995). The long-term effects of parental divorce on the mental health of young adults: A developmental perspective. Child Development, 66, 1614–1634. Chase-Lansdale, P.L., & Gordon, R.A. (1996). Economic hardship and the development of five- and six-year-olds: Neighborhood and regional perspectives. Child Development, 67, 3338–3367. Chisholm, K. (1998). A three-year follow-up of attachment and indiscriminate friendliness in children adopted from Romanian orphanages. Child Development, 69, 1092–1106. Chisholm, K., Carter, M., Ames, E.W., & Morison, S.J. (1995). Attachment security and indiscriminately friendly behavior in children adopted from Romanian orphanages. Development and Psychopathology, 7, 283–294. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Conger, J. (1971). A world they never knew: The family and social change. Daedalus 100, 1105–1138. Conger, R., & Elder, G.E. (1994). Families in troubled times: Adapting to change in rural America. New York: Aldine. Conger, R., Ge, X., Elder, G.H., Lorenz, F., & Simons, R. (1994). Economic stress, coercive family process and developmental problems of adolescents. Child Development, 65, 541–561. Cowan, P.A., & Cowan, C.P. (in press). What an intervention design reveals about how parents affect their children’s academic achievement and social competence. In J.Borkowski, S.Landesman-Ramey, & M.Bristol (Eds.), Parenting and the child’s world: Multiple influences on intellectual and social—emotional development. Hillsdale, NJ: Erlbaum. Cowan, P.A., Powell, D., & Cowan, C.P. (1998). Parenting interventions: A family systems perspective. In W.Damon, I.Sigel, & K.A.Renninger (Eds.), Handbook of child psychology: Child psychology in practice (Vol. 4, pp. 3–72). New York: Wiley. Darling, N., & Steinberg, L. (1997). Community influences on adolescent achievement and deviance. In J.Brooks-Gunn, G.Duncan, & L.Aber (Eds.), Neighborhood poverty: Context and consequences for children: Conceptual, methodological, and policy approaches to studying neighborhoods (Vol. 2, pp. 120–131). New York: Russell Sage Foundation. DeBaryshe, B., Patterson, G., & Capaldi, D. (1993). A performance model for academic achievement in early adolescent boys. Developmental Psychology, 29, 795–804. Devereux, E.C. (1970). The role of peer group experience in moral development. In J. P.Hill (Ed.), Minnesota Symposia on Child Psychology (Vol. 4, pp. 94–140). Minneapolis: University of Minnesota Press.
Contemporary research on parenting 133 Dishion, T., Patterson, G., Stoolmiller, M., & Skinner, M. (1991). Family, school, and behavioral antecedents to early adolescent involvement with antisocial peers. Developmental Psychology, 27, 172–180. Downey, G., & Walker, E. (1992). Distinguishing family-level and child-level influences on the development of depression and aggression. Development and Psychopathology, 4, 81–96. Dunn, J., & Plomin, R. (1986). Determinants of maternal behavior toward three-yearold siblings. British Journal of Developmental Psychology, 57, 348–356. Eisenberg, N., & Damon, W. (Eds.). (1998). Handbook of child psychology: Social, emotional, and personality development (Vol. 3). New York: Wiley. Emde, R., Plomin, R., Robinson, J., Corley, R., DeFries, J., Fulker, D., Reznick, J.S., Campos, J., Kagan, J., & Zahn-Waxler, C. (1992). Temperament, emotion, and cognition at fourteen months: The Mac Arthur longitudinal twin study. Child Development, 63, 1437–1455. Epstein, J.L. (1983). The influence of friends on achievement and affective outcomes. In J.L.Epstein & N.Karweit (Eds.), Friends in school (pp. 177–200). New York: Academic Press. Fletcher, A., Darling, N., Steinberg, L., & Dornbusch, S. (1995). The company they keep: Relation of adolescents’ adjustment and behavior to their friends’ perceptions of authoritative parenting in the social network. Developmental Psychology, 31, 300–310. Forehand, R., & King, H.E. (1977). Noncompliant children: Effects of parent training on behavior and attitude change. Behavior Modification, 1, 93–108. Forehand, R., Wells, K.C., & Griest, D.L. (1980). An examination of the social validity of a parent training program. Behavior Therapy, 11, 488–502. Forgatch, M.S. (1991). The clinical science vortex: A developing theory of antisocial behavior. In D.Pepler & K.Rubin (Eds.), The development and treatment of childhood aggression (pp. 291–315). Hillsdale, NJ: Erlbaum. Forgatch, M.S., & DeGarmo, D.S. (1999). Parenting through change: An effective prevention program for single mothers. Journal of Consulting and Clinical Psychology, 67, 711–724. Fuligni, A., & Eccles, J. (1993). Perceived parent-child relationships and early adolescents’ orientation toward peers. Developmental Psychology, 29, 622–632. Furstenberg, F., Jr., Eccles, J., Elder, G., Jr., Cook, T., & Sameroff, A. (1997). Managing to make it. Chicago: University of Chicago Press. Gatz, M., Pedersen, N.L., Plomin, R., Nesselroade, J.R., & McLearn, G.E. (1992). Importance of shared genes and shared environments for symptoms of depression in older adults. Journal of Abnormal Psychology, 101, 701–708. Ge, X., Conger, R., Cadoret, R., Neiderhiser, J., Yates, W., Troughton, E., & Stewart, M. (1996). The developmental interface between nature and nurture: A mutual influence model of child antisocial behavior and parent behavior. Developmental Psychology, 32, 547–598. Ge, X., Conger, R., Lorenz, F., Shanahan, M., & Elder, G. (1995). Mutual influences in parent and adolescent psychological distress. Developmental Psychology, 31, 406–419. Ge, X., Lorenz, F., Conger, R., Elder, G., & Simons, R. (1994). Trajectories of stressful life events and depressive symptoms during adolescence. Developmental Psychology,
Annual progress in child psychiatry and child development 2000–2001
134
30, 467–483. Gesell, A., & Thompson, H. (1934). Infant behavior: Its genesis and growth. New York: McGraw-Hill. Ghodsion-Carpey, J., & Baker, L.A. (1997). Genetic and environmental influences on aggression in 4- to 7-year-old twins. Aggressive Behavior, 13, 173–186. Goldsmith, H., & Alansky, J. (1987). Maternal and infant temperamental predictors of attachment: A meta-analytic review. Journal of Consulting and Clinical Psychology, 55, 805–816. Goodman, R. (1991). Growing together and growing apart: The nongenetic forces on children in the same family. In R.McGuffin & R.Murry (Eds.), The new genetics of mental illness (pp. 212–224). Oxford, England: Oxford University Press. Greenough, W., & Black, J. (1992). Induction of brain structure by experience: Substrates for cognitive development. In M.R.Gunnar & C.A.Nelson (Eds.), Developmental neuroscience: Minnesota Symposia on Child Psychology (Vol. 24, pp. 155–200). Hillsdale, NJ: Erlbaum. Grossmann, K., Grossman, K., Spangler, G., Suess, G., & Unzner, L. (1985). Maternal sensitivity and newborns’ orientation responses as related to quality of attachment in Northern Germany. Monographs of the Society for Research in Child Development, 50 (1–2, Serial No. 209), 233–256. Gunnar, M. (in press). Early adversity and the development of stress reactivity and regulation. In C.A.Nelson (Ed.), The effects of adversity on neurobehavioral development: Minnesota Symposia on Child Psychology (Vol. 31). Mahwah, NJ: Erlbaum. Hagekull, B. (1989). Longitudinal stability of temperament within a behavioral style framework. In G.A.Kohnstamm, J.E.Bates, & M.K.Rothbart (Eds.), Temperament in childhood (pp. 283–297). Chichester, England: Wiley. Hagekull, B. (1994). Infant temperament and early childhood functioning: Possible relations to the five-factor model. In C.J.Halverson, Jr., G.A.Kohnstamm, &R. P.Martin (Eds.), The developing structure of temperament and personality (pp. 227– 240). Hillsdale, NJ: Erlbaum. Hardy-Brown, K. (1983). Universals in individual differences: Disentangling two approaches to the study of language acquisition. Developmental Psychology, 19, 610– 624. Hardy-Brown, K., & Plomin, R. (1985). Infant communicative development: Evidence from adoptive and biological families for genetic and environmental influences on rate differences. Developmental Psychology, 27, 378–385. Hardy-Brown, K., Plomin, R., & DeFries, J.C. (1981). Genetic and environmental influences on rate of communicative development in the first year of life. Developmental Psychology, 17, 704–717. Harris, J.R. (1995). Where is the child’s environment? A group socialization theory of development. Psychological Review, 102, 458–489. Harris, J.R. (1998). The nurture assumption: Why children turn out the way they do. New York: Free Press. Hartup, W.W. (1989). Social relationships and their developmental significance. American Psychologist, 44, 120–126. Hetherington, E.M. (1989). Coping with family transitions: Winners, losers, and
Contemporary research on parenting 135 survivors. Child Development, 60, 1–14. Hetherington, E.M. (1991). The role of individual differences in family relations in coping with divorce and remarriage. In P.Cowan & E.M.Hetherington (Eds.), Advances in family research: Family transitions (Vol. 2, pp. 165–194). Hillsdale, NJ: Erlbaum. Hetherington, E.M. (1993). A review of the Virginia Longitudinal Study of Divorce and Remarriage: A focus on early adolescence. Journal of Family Psychology, 7, 39–56. Hoffman, L.W. (1991). The influence of the family environment on personality: Accounting for sibling differences. Psychological Bulletin, 110, 187–203. Hogue, A., & Steinberg, L. (1995). Homophily of internalized distress in adolescent peer groups . Developmental Psychology, 31, 897–906. Holden, G.W., & Miller, P.C. (1999). Enduring and different: A meta-analysis of the similarity in parents’ child rearing. Psychological Bulletin, 125, 223–254. Huttenlocher, P.R. (1994). Synaptogenesis, synapse elimination, and neural plasticity in human cerebral cortex. In C.A.Nelson (Eds.), Minnesota Symposia on Child Psychology: Threats to optimal development: Integrating biological, psychological, and social risk factors (Vol. 27, pp. 35–54). Hillsdale, NJ: Erlbaum. Kandel, D. (1978). Homophily, selection, and socialization in adolescent friendships. American Journal of Sociology, 84, 427–436. Kim, J.E., Hetherington, E.M., & Reiss, D. (1999). Associations between family relationships, antisocial peers and adolescent’s externalizing behaviors: Gender and family type differences. Child Development, 70, 1209–1230. Klebanov, P.K., Brooks-Gunn, J., & Duncan, G.T. (1994). Does neighborhood and family poverty affect mothers’ parenting, mental health and social support? Journal of Marriage and the Family, 56, 441–455. Kochanska, G. (1995). Children’s temperament, mothers’ discipline, and the security of attachment: Multiple pathways to emerging internalization . Child Development, 66, 597–615. Kochanska, G. (1997). Multiple pathways to conscience for children with different temperaments: From toddlerhood to age 5. Developmental Psychology, 33, 228–240. Kuczynski, L. (1984). Socialization goals and mother-child interaction: Strategies for long-term and short-term compliance. Developmental Psychology, 20, 1061–1073. Lee, C.L., & Bates, J. (1985). Mother-child interaction at age two years and perceived difficult temperament. Child Development, 56, 1314–1325. Lewis, M. (1997). Altering fate: Why the past does not predict the future. New York: Guilford. Liu, D., Diorio, J., Tannenbaum, B., Caldji, C., Francis, D., Freedman, M.A., Sharma, S., Pearson, P., Plotsky, P.M., & Meaney, M.J. (1997, September 12). Maternal care, hippocampal glucocorticoid receptors and hypothalamic-pituitary-adrenal responses to stress. Science, 277, 1659–1662. Maccoby, E.E. (1992). The role of parents in the socialization of children: An historical overview. Developmental Psychology, 28, 1006–1017. McCall, R. (1991). So many interactions, so little evidence: Why? In T.Wachs & R. Plomin (Eds.), Conceptualization and measurement of organism—environment interaction (pp. 142–161). Washington, DC: American Psychological Association.
Annual progress in child psychiatry and child development 2000–2001
136
McCartney, K., Harris, M., & Bernieri, F. (1990). Growing up and growing apart: A developmental meta-analysis of twin studies. Psychological Bulletin, 107, 226–237. McClelland, G., & Judd, C. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin, 114, 376–390. McGue, M. (1994). Genes, environment, and the etiology of alcoholism. In R.Zucker, G.Boyd, & J.Howard (Eds.), The development of alcohol problems: Exploring the biopsychosocial matrix of risk (National Institute of Alcohol Abuse and Alcoholism Research Monograph No. 26, pp. 1–40). Rockville, MD: U.S. Department of Health and Human Services. McLoyd, V. (1990). The impact of economic hardship on Black families and children: Psychological distress, parenting, and socioemotional development. Child Development, 61, 311–346. McMahon, R.J., & Wells, K.C. (1998). Conduct problems. In E.J.Mash & R.A. Barkley (Eds.), Treatment of childhood disorders (2nd ed., pp. 111–151). New York: Guilford. Miles, D., & Carey, G. (1997). Genetic and environmental architecture of human aggression. Journal of Personality and Social Psychology, 72, 207–217. Miyake, K., Chen, S. -J., & Campos, J. (1985). Infant temperament, mother’s mode of interaction, and attachment in Japan: An interim report. Monographs of the Society for Research in Child Development, 50 (1–2, Serial No. 209), 276–297. Mounts, N., & Steinberg, L. (1995). An ecological analysis of peer influence on adolescent grade point average and drug use. Developmental Psychology, 31, 915–922. Nelson, C.A. (in press). The neurobiological bases of early intervention. In S.J.Meisels & J.P.Shonkoff (Eds.), Handbook of early childhood intervention (2nd ed.). New York: Cambridge University Press. O’Connor, T.G., Deater-Deckard, K., Fulker, D., Rutter, M.L., & Plomin, R. (1998). Genotype-environment correlations in late childhood and early adolescence: Antisocial behavioral problems and coercive parenting. Developmental Psychology, 34, 970–981. Parke, R., & Bhavnagri, N.P. (1989). Parents as managers of children’s peer relationships. In D.Belle (Ed.), Children’s social networks and social support (pp. 241–259). New York: Wiley. Patterson, G.R. (1975). Multiple evaluations of a parent-training program. In T.Thompson & W.S.Dockens (Eds.), Applications of behavior modification (pp. 299– 322). New York: Academic Press. Patterson, G.R. (1999). Recent news concerning the demise of parenting may be a bit premature. Unpublished manuscript, Oregon Social Learning Center, Eugene, OR. Patterson, G.R., DeBaryshe, B.D., & Ramsey, E. (1989). A developmental perspective on antisocial behavior. American Psychologist, 44, 329–335. Plomin, R. (1990). Nature and nurture: An introduction to human behavioral genetics. Pacific Grove, CA: Brooks/Cole. Plomin, R., & Daniels, D. (1987). Why are children in the same family so different from each other? Behavioral and Brain Sciences, 10, 1–16. Plomin, R., DeFries, J., & Fulker, D. (1988). Nature and nurture during infancy and early childhood. New York: Cambridge University Press. Plomin, R., & Rutter, M. (1998). Child development, molecular genetics, and what to do with genes once they are found. Child Development, 69, 1223–1242.
Contemporary research on parenting 137 Quinton, D., Pickles, A., Maughan, B., & Rutter, M. (1993). Partners, peers, and pathways: Assortative pairing and continuities in conduct disorder. Development and Psychopathology, 5, 763–783. Reiss, D., Niederhiser, J., Hetherington, E.M., & Plomin, R. (in press). The relationship code: Deciphering genetic and social patterns in adolescent development. Cambridge, MA: Harvard University Press. Riksen-Walraven, J. (1978). Effects of caregiver behavior on habituation rate and selfefficacy in infants. International Journal of Behavioral Development, 1, 105–130. Rose, R. (1995). Genes and human behavior. Annual Review of Psychology, 46, 625–654. Rothbart, M., & Ahadi, S. (1994). Temperament and the development of personality. Journal of Abnormal Psychology, 103, 55–66. Rothbart, M., & Bates, J. (1998). Temperament. In W.Damon & N.Eisenberg (Eds.), Handbook of child psychology: Social, emotional, and personality development (Vol. 3, pp. 105–176). New York: Wiley. Rowe, D. (1983). A biometrical analysis of perceptions of family environment: A study of twin and singleton sibling kinship. Child Development, 54, 416–423. Rowe, D. (1994). The limits of family influence: Genes, experience, and behavior. New York: Guilford. Rowe, D., Chassin, L., Presson, C., Edwards, D., & Sherman, S.J. (1992). An “epidemic” model of adolescent cigarette smoking. Journal of Applied Social Psychology, 22, 261–285. Rutter, M. (1990). Psychosocial resilience and protective mechanisms. In J.Rolf, A. S.Masten, D.Cicchetti, K.H.Nuechterlein, & S.Weintraub (Eds.), Risk and protective factors in the development of psychopathology (pp. 181–214). New York: Cambridge University Press. Rutter, M., Dunn, J., Plomin, R., Simonoff, E., Pickles, A., Maughan, B., Ormel, H., Meyer, J., & Eaves, L. (1997). Integrating nature and nurture: Implications of personenvironment correlations and interactions for developmental psychopathology. Development and Psychopathology, 9, 335–364. Rutter, M., & the English and Romanian Adoptees (ERA) study team. (1998). Developmental catch-up, and deficit, following adoption after severe global early privation. Journal of Child Psychology and Psychiatry and Allied Disciplines, 39, 465– 476. Rutter, M., & Quinton, D. (1984). Parental psychiatric disorder: Effects on children. Psychological Medicine, 14, 853–880. Scarr, S., & Weinberg, R.A. (1976). IQ test performance of Black children adopted by White families. American Psychologist, 31, 726–739. Scarr, S., & Weinberg, R.A. (1978). The influence of “family background” on intellectual attainment. American Sociological Review, 43, 674–692. Schiff, M., Duyme, M., Dumaret, A., & Tomkiewitz, S. (1982). How much could we boost scholastic achievement and IQ scores? A direct answer from a French adoption study. Cognition, 12, 165–196. Skodak, M., & Skeels, H. (1949). A final follow-up of one hundred adopted children. Journal of Genetic Psychology, 75, 85–125. Steinberg, L., Lamborn, S., Darling, N., Mounts, N., & Dornbusch, S. (1994). Overtime
Annual progress in child psychiatry and child development 2000–2001
138
changes in adjustment and competence among adolescents from authoritative, authoritarian, indulgent, and neglectful families. Child Development, 65, 754–770. Stern, D.N. (1985). The interpersonal world of the infant. New York: Basic Books. Stoolmiller, M. (1999). Implications of the restricted range of family environments for estimates of heritability and nonshared environment in behavior-genetic adoption studies. Psychological Bulletin, 125, 392–409. Suomi, S.J. (1997). Long-term effects of different early rearing experiences on social, emotional and physiological development in nonhuman primates. In M.S. Kesheven & R.M.Murra (Eds.), Neurodevelopmental models of adult psychopathology (pp. 104– 116). Cambridge, UK: Cambridge University Press. Suomi, S.J. (in press). A biobehavioral perspective on developmental psychopathology: Excessive aggression and serotonergic dysfunction in monkeys. In A.J. Sameroff, M.Lewis, & S.Miller (Eds.), Handbook of developmental psychopathology. New York: Plenum. Tejerina-Allen, M., Wagner, B.M., & Cohen, P. (1994). A comparison of across-family and within-family parenting predictors of adolescent psychopathology and suicidal ideation. In E.M.Hetherington, D.Reiss, & R.Plomin (Eds.), Separate social worlds of siblings: The impact of nonshared environment on development (pp. 143–158). Hillsdale, NJ: Erlbaum. Tienari, P., Wynne, L.C., Moring, J., Lahti, I., Naarala, M., Sorri, A., Wahlberg, K.-E., Saarento, O., Seitma, M., Kaleva, M., & Lasky, K. (1994). The Finnish adoptive family study of schizophrenia: Implications for family research. British Journal of Psychiatry, 23(Suppl. 164), 20–26. Turkheimer, E., & Waldron, M.C. (in press). Nonshared environment: A theoretical, methodological, and quantitative review. Psychological Bulletin. Van den Boom, D.C. (1989). Neonatal irritability and the development of attachment. In G.A.Kohnstamm, J.E.Bates, & M.K.Rothbart (Eds.), Temperament in childhood (pp. 299–318). Chichester, UK: Wiley. Van den Boom, D.C. (1994). The influence of temperament and mothering on attachment and exploration: An experimental manipulation of sensitive responsiveness among lower-class mothers with irritable infants. Child Development, 65, 1457–1477. Wachs, T.D. (1992). The nature of nurture. Newbury Park, CA: Sage. Wachs, T.D. (1999). Celebrating complexity: Conceptualization and assessment of the environment. In S.Friedman & T.D.Wachs (Eds.), Measuring environment across the life span: Emerging methods and concepts (pp. 357–392). Washington, DC: American Psychological Association. Wachs, T.D., & Plomin, R. (1991). Conceptualization and measurement of organismenvironment interaction. Washington, DC: American Psychological Association. Wahlsten, D. (1990). Insensitivity of the analysis of variance to heredity-environment interaction. Behavior and Brain Sciences, 13, 109–161. Weinberg, R.A. (1989). Intelligence and IQ: Landmark issues and great debates. American Psychologist, 44, 98–104. Werner, E. (1990). Protective factors and individual resilience. In S.Meisels & J. Shonkoff (Eds.), Handbook of early childhood intervention (pp. 97–116). Cambridge, MA: Harvard University Press.
Contemporary research on parenting 139 Werner, E., & Smith, R. (1992). Overcoming the odds: High risk children from birth to adulthood. Ithaca, NY: Cornell University Press. Zill, N., Morrison, D., & Coiro, M. (1993). Long-term effects of parental divorce on parent-child relationships, adjustment, and achievement in young adulthood. Journal of Family Psychology, 7, 91–103.
PART II: PARENTING
8 Social versus Biological Parenting: Family Functioning and the Socioemotional Development of Children Conceived by Egg or Sperm Donation Susan Golombok, Clare Murray, Peter Brinsden, and Hossam Abdalla
By investigating egg donation families, donor insemination families, adoptive families, and families created by in vitro fertilization, the aim of the present study was to examine parents’ emotional well being, the quality of parenting, and childrens’ socioemotional development in families with a child who is genetically unrelated to the mother or the father. The differences that were found to exist between families according to the presence or absence of genetic ties between parents and their children reflected greater psychological well being among mothers and fathers in families where there was no genetic link between the mother and the child. The families did not differ with respect to the quality of parenting or the psychological adjustment of the child.
INTRODUCTION Although the use of donor sperm to enable couples with an infertile male partner to have children has been practiced for many years, it is only since 1983, following advances in reproductive technology, that infertile women have been able to conceive a child using a donated egg (Lutjen et al., 1984; Trouson, Leeton, Besanka, Wood, & Conti, 1983). This procedure involves fertilization of the donated egg with the father’s sperm in the laboratory, followed by the transfer of the resulting embryo to the mother’s uterus. Thus, it is now possible for children to be born to, and raised by, mothers with whom they have no genetic link. A number of concerns have been expressed regarding the potential negative consequences of gamete donation for children’s psychological well being, the most common of which is that the practice of keeping information about genetic origins secret from the child may have an adverse effect on the quality of parent-child relationships and consequently on the child (Daniels & Taylor, 1993; Schaffer & Diamond, 1993). As few children are told that a donated sperm or egg had been used in their conception, the large
Annual progress in child psychiatry and child development 2000–2001
142
majority grow up not knowing that their father or mother is genetically unrelated to them. Findings suggestive of an association between secrecy about genetic parentage and negative outcomes for children have come from research on adoption. It has been demonstrated that adopted children benefit from knowledge about their biological parents, and that children who are not given such information may become confused about their identity and at risk for emotional problems (Hoopes, 1990; Sants, 1964; Schechter & Bertocci, 1990; Triseliotis, 1973). In the field of assisted reproduction, parallels have been drawn with the adoptive situation and it has been suggested that lack of knowledge of, or information about, the donor may be harmful for the child (Clamar, 1989; Snowden, 1990; Snowden, Mitchell, & Snowden, 1983). From a family therapy perspective, secrets are believed to be detrimental to family functioning because they create boundaries between those who know and those who do not, and cause anxiety when topics related to the secret are discussed (Karpel, 1980). In examining the particular case of parents keeping secrets from their children, Papp (1993) argued that children can sense when information is being withheld due to the taboo that surrounds the discussion of certain topics, and that they may become confused and anxious, or even develop symptoms of psychological disorder, as a result. A further concern raised by the use of gamete donation is that parents may feel or behave less positively toward a nongenetic than a genetic child. It has been argued that the child may not be fully accepted as part of the family, and that the absence of a genetic tie to one or both parents may have an undermining effect on the child’s sense of identity (Burns, 1987). It has also been suggested that whether or not gamete donation has been used in the child’s conception, the stress of infertility may lead to dysfunctional patterns of parenting, which may result in negative outcomes for the child (Burns, 1990). In spite of the expectations that children conceived by gamete donation may be at risk for psychological problems, a previous study of assisted reproduction families by the present authors (Golombok, Cook, Bish, & Murray, 1995) found a greater involvement in parenting among donor insemination parents than among a control group of parents with a naturally conceived child, with no differences in the quality of parent-child relationships between donor insemination parents and either adoptive parents or parents with a genetically related child conceived by in vitro fertilization. The children in these different family types were functioning well and did not differ with respect to their emotions, behavior, or relationships. It was concluded that a strong desire for parenthood seemed to be more important than genetic relatedness for fostering positive family relationships, and that conception by sperm donation did not appear to have an adverse effect on the socioemotional development of the child. In the current investigation, an additional group of families with a child conceived by egg donation was recruited, and examined together with the donor insemination, adoptive, and in vitro fertilization (IVF) families from the previous study (Golombok et al., 1995). The comparison between families according to whether or not the child was genetically related to the mother (donor insemination and IVF families vs. egg donation and adoptive families), and whether or not the child was genetically related to the father (egg donation and IVF families versus donor insemination and adoptive families), as well as the interaction between them, provides an opportunity to examine the consequences of social versus biological parenting in families matched for the parents’ strong desire for a
Social versus biological parenting 143 child. Unlike the families with a naturally conceived child in the previous investigation, all four groups of families had experienced a period of infertility preceding the transition to parenthood, and had thus been highly committed to raising a child. From the findings of the previous study it is expected that egg donation parents, like donor insemination parents, would have positive relationships with their much-wanted children and that the children themselves would be functioning well. It is conceivable, however, that differences may exist with respect to the quality of parenting and children’s socioemotional development according to whether it is the mother or the father who lacks a genetic link with the child. There is a growing body of empirical evidence to show that the course of a child’s social and emotional development is related to the quality of the child’s attachment to parents (Bowlby, 1969, 1973; Main, Kaplan, & Cassidy, 1985), and that secure attachment relationships are fostered through parents’ sensitive responding to the child (Grossmann, Grossmann, Spangler, Suess, & Unzer, 1985; Isabella, Belsky, & von Eye, 1989). Other aspects of parenting have also been shown to influence children’s psychological well being. For example, Baumrind (1989) has demonstrated that an authoritative parenting style, i.e. a combination of warmth and discipline, has positive outcomes for children’s socioemotional development. As mothers generally interact more with their children than do fathers (Lamb, 1997; Parke, 1995), and their identity is more tied up with their parental role than it is for fathers (Hoffman, Thornton, & Marris, 1978; Woollett, 1991), the absence of a genetic link with the child may matter less for mothers than fathers in terms of the quality of parent-child relationships and outcomes for the child. In addition, fathers are more concerned than mothers about the need to achieve immortality through their offspring (Hoffman et al., 1978), which may cause them to be less committed to nongenetic children than to children with whom they share a genetic link. Thus, differences between families according to genetic relatedness between the mother and the child would not necessarily be predicted, whereas more nega-tive outcomes may be expected in families where the child and the father are genetically unrelated compared with families where a genetic link exists between the father and the child.
METHOD Participants Forty-one families with a child conceived by IVF (28 boys and 13 girls), 45 families with a child conceived by donor insemination (25 boys and 20 girls), and 21 families with a child conceived by egg donation (13 boys and eight girls) were obtained through infertility clinics throughout the United Kingdom. Total populations of IVF and donor insemination families with a child aged between 4 to 8 years from each participating clinic were asked to take part in the research. For the egg donation families, the lower age limit was set at 3½ years in order to increase the sample size of this relatively new family type. The donor was anonymous to all of the donor insemination families and 18 of the egg donation families. In the case of the three egg donation families who conceived their child with the help of a known donor, the donor was a family friend. The response rate
Annual progress in child psychiatry and child development 2000–2001
144
for IVF, donor insemination, and egg donation families respectively was 95%, 62%, and 81%.1 Fifty-five adoptive families were recruited through adoption agencies by approaching families with a 4- to 8-year-old child who had been adopted in the first 6 months of life (28 boys and 27 girls). The adopted children were born in the United Kingdom and were of the same ethnic origin as their adoptive parents. The response rate for adoptive families was 76%. Children with major congenital abnormalities, children who had experienced obstetric or perinatal complications that were thought likely to involve brain damage or risk of persisting disability, and children of a multiple birth were not included in the study. There were similar proportions of boys and girls in each group of families. A significant difference between groups was found for age of the child [F(3, 158)=20.16, p<0.0001]. The adopted children were the oldest, aged 6 years 3 months on average, and the egg donation children the youngest, with a mean age of 4 years 6 months. Similarly, a significant group difference was found for the age of the mothers [F(3, 158)=10.32, p<0.0001]. In this case, the egg donation mothers were the oldest (mean age 41 years) and the donor insemination mothers were the youngest (mean age 36 years). Social class was rated according to the occupation of the parent with the highest ranking position (1— professional/managerial occupations; 2—skilled nonmanual occupations; 3—skilled manual occupations; 4—partly skilled/unskilled occupations). There was a significant difference between groups for social class [F(3, 158)=10.77, p<0.0001], with the egg donation families receiving the highest ratings and the donor insemination families the lowest. The families were predominantly white. No significant difference between groups was identified for the number of children in the family, and a higher proportion of target children in each family type was first-born than later-born, except for the adoptive children, where there was an equal proportion of first-born and later-born children. Almost all of the parents were married (one egg donation couple was cohabiting), and only six sets of parents (3.8%) had separated or divorced (two IVF, two donor insemination, one adoptive, and one egg donation). As significant group differences were found for age of the child, age of the mother, and social class, these demographic variables were entered into all of the analyses as covariates. All of the parents were contacted in the first instance by a letter from the clinic or adoption agency. Those who agreed to participate were visited at home by a researcher trained in the study techniques. Data were collected from the mother by interview, from both parents by questionnaire, and from the child using a standardized test. Measures Parents’ Marital and Psychological State. Both the mother and the father completed the Golombok Rust Inventory of Marital State (Rust, Bennun, Crowe, & Golombok, 1988; Rust, Bennun, & Golombok, 1990), a questionnaire measure of the quality of the marital relationship. Scores range from 0 to 84 with a score of 34 or above indicating marital dissatisfaction. The Beck Depression Inventory (Beck & Steer, 1987; Steer, Beck, & 1. The response rate for egg donation families is estimated as one clinic did not keep an accurate record of those who declined to take part.
Social versus biological parenting 145 Garrison, 1986) and the Trait Anxiety Inventory (Spielberger, 1983) were also completed by both parents to assess depression and anxiety respectively. Scores range from 0 to 63 for the Beck Depression Inventory with a score of 10 or above indicating mild-moderate depression, and from 20 to 80 for the Trait Anxiety Inventory, with 35 representing the average score for working adults aged between 19 and 39 years. All three of these instruments have been shown to have good reliability and to discriminate well between clinical and nonclinical groups. The short form of the Parenting Stress Index (PSI/SF; Abidin, 1990) was also administered to both parents to provide a standardized assessment of stress associated with parenting for mothers and fathers separately. This measure produces a total score of the overall level of parenting stress an individual is experiencing, with a score of 86 or above indicating a clinically significant level of stress, as well as the four subscale scores of parental distress (feelings of parental incompetence, stresses associated with restrictions on lifestyle, conflict with the child’s other parent, lack of social support, and depression), parent-child dysfunctional interaction (the parent’s perception that the child does not measure up to expectations and that interactions with the child are not reinforcing), difficult child (the behavioral characteristics of children that make them easy or difficult to manage), and defensive responding (social desirability). Test-retest reliability for this instrument has been shown to be high over a 6-month period. Concurrent and predictive validity has been demonstrated for the full-length questionnaire, and the short form has been reported to correlate very highly with the fulllength version. Quality of Parenting. The quality of parenting was assessed by standardized interview with the mother using an adaptation of the technique developed by Quinton and Rutter (1988). This procedure has been validated against observational ratings of mother-child relationships in the home, demonstrating a high level of agreement between global ratings of the quality of parenting by interviewers and observers (concurrent validity, r=.63). The researchers were fully trained in the interview techniques by one of the authors of the interview procedure (DQ). The interview, which was tape-recorded, lasted for around 1½ hours and was conducted with the mother alone. Detailed accounts were obtained of the child’s behavior and the parents’ response to it. The mothers were asked to describe the child’s daily routine focusing on waking, meal times, leaving for school/day care, returning home, mother’s and father’s play activities with the child, and bedtime. Information was obtained on the parents’ handling of any problems associated with these areas, and particular attention was paid to parent-child interactions relating to issues of control and the child’s fears and anxieties. Four overall ratings of the quality of parenting were made taking into account information obtained from the entire interview: 1) warmth was rated on a 6-point scale ranging from 0, “none,” to 5, “high.” This rating of the mother’s warmth toward the child was based upon the mother’s tone of voice and facial expression when talking about the child, spontaneous expressions of warmth, sympathy, and concern about any difficulties experienced by the child, and enthusiasm and interest in the child as a person; 2) emotional involvement was rated on a 5-point scale from 0, “little or none,” to 4, “extreme.” This rating, which represented anxious over-involvement at the extreme end, took account of the extent to which the family day was organized around the child, the
Annual progress in child psychiatry and child development 2000–2001
146
extent to which the needs or interests of the child were placed before those of other family members, the extent to which the mother was over-concerned, over-protective, or inhibited the child from age-appropriate independent activities, the extent to which the mother was willing to leave the child with other caretakers, and the extent to which the mother had interests or engaged in activities apart from those relating to the child; 3) mother-child interaction; and 4) father-child interaction were each rated on a 5-point scale ranging from 0, “very low,” to 4, “very high.” These ratings of the quality of interaction be-tween the parent and the child were based upon mothers’ reports of the extent to which the parent and the child enjoyed each other’s company, wanted to be with each other, spent time together, enjoyed joint play activities, and showed physical affection to one another. Interview data also were used to make ratings of the father’s contribution to parenting with respect to helping the mother with childrearing (rated on a 5-point scale from 0, “no help,” to 4, “takes a major parenting load”) and parental coordination over control (rated on a 5point scale from 0, “active uncoordination,” to 4, “coordinated action”). In addition, the seriousness of disputes between the mother and the child was rated on a 4-point scale from 0, “minor episodes,” to 3, “major battles.” Although the validity of the mothers’ reports of father-child interaction has not been established using observational ratings of father-child relationships, a correlation of .4 was found in the previous study by Golombok et al. (1995) between mothers’ reports of father-child interaction and fathers’ reports of the child being difficult to manage as measured by the difficult child subscale of the PSI/SF. This gives some evidence for the validity of the mothers’ reports of father-child interaction, particularly in view of the differences between these two constructs. In the previous study, 27 randomly selected interviews were coded by a second interviewer who was “blind” to family type in order to calculate inter-rater reliabilities. Person product-moment coefficients for warmth, emotional involvement, mother-child interaction, and father-child interaction were found to be 0.75, 0.63, 0.72, and 0.69, respectively. In the present investigation this process was repeated for 18 of the egg donation families (3 were not included due to the poor quality of the tape recordings). The reliability coefficients for warmth, emotional involvement, mother—child interaction, father-child interaction, father’s contribution to parenting, parental coordination over control, and seriousness of disputes between the mother and the child were 0.66, 0.51, 0.46, 0.74, 0.77, 0.77, and 0.89 respectively. In addition, mothers of children conceived by egg donation or donor insemination were interviewed about their openness regarding the circumstances of their child’s conception. Systematic information was obtained from these mothers with respect to whether or not they had told their child about his or her genetic origins, and whether or not they had told the child’s grandparents or family friends. Those who did not plan to tell their child were asked about their reasons for secrecy, and each of the following variables was coded as “yes” or “no” according to the mother’s responses: 1) to protect the child; 2) to protect the father; 3) to protect the mother; and 4) there is no need to tell. Children’s Socioemotional Development. The presence of behavioral and emotional problems in the child was assessed using the Rutter “A” scale, which is completed by the mother and produces an overall score of the child’s psychiatric state. This questionnaire has been shown to have good inter-rater and testretest reliability, and to discriminate well between children with and without psychiatric disorder (Rutter, Cox, Tupling, Berger, &
Social versus biological parenting 147 Yule, 1975; Rutter, Tizard, &Whitmore, 1970). Each child was administered the Pictorial Scale of Perceived Competence and Social Acceptance for Young Children (Harter & Pike, 1984). This is a measure of children’s perceptions of their cognitive and physical competencies, and of their perceptions of acceptance by their mother and by peers, all of which have been shown to be associated with the development of self-esteem in later childhood. Children’s perceptions in these domains do not necessarily reflect their actual competencies or acceptance by others. A score is obtained for each of the following subscales; 1) cognitive competence, 2) physical competence, 3) maternal acceptance, and 4) peer acceptance. The higher the score, the more positive the child’s feelings of competence and social acceptance. Satisfactory internal consistency has been demonstrated, with coefficient alpha values ranging from .85 to .89 for the different age groups of children studied. The scale has been shown to discriminate between groups of children in predicted ways, for example, between peer acceptance and length of time at a school, and between perceived cognitive competence and academic achievement at school, indicating that it is a valid measure.
RESULTS Parents’ Marital and Psychiatric State Table 8.1 shows the means, standard deviations, F values, and significance levels for all of the measures of parents’ marital and psychiatric state by group. Multivariate analysis of variance (MANOVA) was conducted with two betweensubjects factors: (a) whether or not the child was genetically related to the mother (donor insemination and IVF families vs. egg donation and adoptive families), and (b) whether or not the child was genetically related to the father (egg donation and IVF families vs. donor insemination and adoptive families). The dependent variables were mothers’ and fathers’ total scores on the Golombok Rust Inventory of Marital State, the Trait Anxiety Inventory, the Beck Depression Inventory, and the PSI/SF. There was no significant main effect for either genetic relatedness to the mother or genetic relatedness to the father using Wilk’s criterion for combined ratings. However, a significant interaction was found [F(8, 110) =2.25, p<0.05]. The data for each variable were then analyzed using 2×2 analyses of covariance with age of child, age of mother, and social class as covariates. With respect to the quality of the parents’ marital relationship for those couples who had not separated or divorced, a significant main effect for marital satisfaction was found for mothers as assessed by the Golombok Rust Inventory of Marital Satisfaction [F(1, 135)=4.32, p<0.05], showing that mothers of genetically unrelated children reported greater marital satisfaction than mothers of genetically related children irrespective of whether or not the father had a genetic link with the child. No differences in marital satisfaction were found for fathers. Neither were there significant group differences for mothers or fathers for depression as assessed by the Beck Depression Inventory.
Annual progress in child psychiatry and child development 2000–2001
148
TABLE 8.1 Means, Standard Deviations (SD), and F Values for Comparisons of Parents’ Marital and Psychiatric State between Family Type
Genetic relationship to Donor Egg IVF Insemination Adoptive Donation MotherFatherInteraction Mean SD Mean SD Mean SD Mean SD F F F Marital state Mother 25.912.4 Father 25.9 9.8 Depression Mother 5.5 4.7 Father 5.0 4.6 Anxiety Mother 36.7 8.3 Father 35.9 8.4 Parenting 68.214.5 stress: Mother Parental 24.1 5.7 distress Dysfunctional 18.6 4.8 interaction Difficult child 25.7 6.4 14.9 4.5 Defensive responding Parenting 70.113.3 stress: Father Parental 23.4 5.0 distress Dysfunctional 20.0 5.3 interaction Difficult child 26.7 6.3 Defensive 13.9 3.1 responding *p<0.05; **p<0.01.
26.6 13.4 22.4 9.6 23.011.0 4.32* 0.16 24.7 10.0 23.2 9.1 24.110.0 0.62 0.06 4.4 4.6
4.3 3.4
4.5 4.5 4.5 3.9
3.9 4.1 3.8 3.5
0.06 0.08
1.68 0.04 2.22 0.68
0.68 0.92
35.1 8.4 37.6 8.6 34.8 7.5 0.88 0.72 36.7 9.0 35.1 7.3 34.5 7.2 4.58* 1.30 63.6 17.8 67.213.3 58.116.5 3.14 2.19
1.80 0.25 7.70**
20.5
7.2 21.9 5.1 20.2 6.2 4.13* 0.11
8.93**
18.0
4.5 19.3 4.8 17.3 5.1
0.33 0.99
2.42
24.9 12.5
8.6 25.8 6.0 22.5 7.1 1.22 2.13 4.5 13.3 3.5 11.9 3.7 5.16* 0.00
3.37 7.24**
71.1 18.8 65.813.9 63.314.4 8.54** 1.96
0.30
23.5
7.2 21.9 6.2 21.5 4.8 4.36* 0.45
0.19
19.8
6.1 18.9 4.4 17.0 4.3 7.38** 3.50
2.45
27.3 14.7
8.3 25.0 6.7 25.5 7.6 4.74* 0.72 4.6 13.1 3.8 12.1 2.6 8.13** 2.66
0.08 0.36
With respect to anxiety level as measured by the Trait Anxiety Inventory, a significant main effect was found for fathers [F(1, 121)=4.58, p<0.05], showing that fathers reported lower levels of trait anxiety when the child was genetically unrelated to the mother irrespective of whether or not there was a genetic link with the father. No significant differences in trait anxiety were found for mothers.
Social versus biological parenting 149 A significant main effect in stress associated with parenting as measured by the total PSI/SF score was also found for fathers [F(1, 122)=8.54, p<0.01], again reflecting lower levels of stress among fathers when the child was genetically unrelated to the mother whether or not there was a genetic relationship with the father. Analysis of the fathers’ subscale scores identified significant main effects for parental distress [F(1, 123)=4.36, p<0.05], parent-child dysfunctional interaction [F(1, 122)=7.38, p<0.01], and difficult child [F(1, 122)=4.74, p<0.05], all in the direction of lower levels of stress reported by the father when the mother was not genetically related to the child. For mothers, significant interactions were found for the total PSI/SF score [F(1, 144)= 7.70, p<0.01] and the parenting distress subscale score [F(1, 145)=8.93, p< 0.01], showing that mothers who were not genetically related to their child reported lower levels of parenting stress when the child had a genetic link with the father. With respect to defensive responding, a significant main effect was found for fathers [F(1, 123)=8.13, p<0.01], indicating that in families where the child was not genetically related to the mother, fathers were less likely to give socially desirable responses to the PSI/SF. A significant interaction was found for mothers [F(1, 145)=7.24, p<0.01], reflecting more defensive responding in families where both parents were genetically related to the child. Quality of Parenting Table 8.2 shows the means, standard deviations, F values, and significance levels for all of the measures of quality of parenting by group. MANOVA was conducted with the same two between-subjects factors as before: (a) whether or not the child was genetically related to the mother, and (b) whether or not the child was genetically related to the father. The dependent variables were warmth, emotional involvement, mother-child interaction, father-child interaction, father’s contribution to parenting, parental coordination over control, and seriousness of disputes between the mother and the child. There was no significant main effect for either genetic relatedness to the mother or genetic relatedness to the father using Wilk’s criterion for combined ratings. However, a significant interaction was found [F(7, 122)=3.22, p<0.01]. The data for each variable were then analyzed using 2×2 analyses of covariance with age of child, age of mother, and social class as covariates. A significant difference was found between groups for warmth [F(1, 151)= 6.19, p<0.05], with less warmth expressed by mothers when the child was not genetically related to the father irrespective or whether or not there was a genetic link with the mother. No significant differences were found in the level of mothers’ emotional involvement with the child. Nor were significant differences identified for either mother—child interaction or father-child interaction. The groups did not differ with respect to the fathers’ overall contribution to parenting. However, a significant interaction was identified for parental coordination over control [F (1, 145)=6.06, p<0.05], showing that in families where the child was not genetically related to the mother, parents reported more coordination over discipline of the child when there was a genetic link between the child and the father. In terms of the seriousness of disputes between the mother and the child, a significant interaction was found [F(1,132)= 4.00, p<0.05], indicating that in families where the child did not have a
Annual progress in child psychiatry and child development 2000–2001
150
genetic link with the mother, more severe disputes were reported when there was a genetic link with the father.
Table 8.2 Means, Standard Deviations (SD) and F Values for Comparisons of Quality of Parenting between Family Type
Genetic relationship to
Mother’s warmth Mother’s emotional involvement Mother-child interaction Father-child interaction Father’s contribution to parenting Coordination over control Seriousness of disputes *p<0.05.
Donor Egg IVF Insemination Adoptive Donation Mother Father Interaction MeanSD Mean SD Mean SD Mean SD F F F 4.10.8 3.8 0.9 3.6 1.1 4.2 0.7 0.02 6.19* 2.41 2.20.5
2.4 0.7
2.1 0.6
2.1 0.8
3.14 0.02
2.70
3.20.5
3.3 0.6
3.0 0.6
3.4 0.5
1.46 0.52
2.38
2.90.7
2.8 0.9
2.8 0.7
3.0 0.8
0.07 1.01
0.11
5.51.0
5.1 1.0
5.3 0.9
5.2 1.1
0.53 0.56
0.83
2.50.9
2.8 0.8
2.9 0.7
3.4 0.5 4.18* 0.04
6.06*
1.00.4
1.2 0.5
0.9 0.4
1.3 0.7
4.00*
0.01 0.01
Children’s Socioemotional Development Table 8.3 shows the means, standard deviations, F values, and significance levels for all of the measures of children’s socioemotional development by group. MANOVA was conducted with the same two between-subjects factors as before: (a) whether or not the child was genetically related to the mother; and (b) whether or not the child was genetically related to the father. The dependent variables were the total “A” scale score and the four subscale scores of the Pictorial Scale of Perceived Competence and Social Acceptance for Young Children; cognitive competence, physical competence, maternal acceptance, and peer acceptance. There was no significant main effect for genetic relatedness to the mother. However, a significant effect was found for genetic relatedness to the father using Wilk’s criterion for combined ratings [F(5, 120)= 2.83. p<0.05]. The interaction was not significant. The data for each variable were then analyzed using 2×2 analyses of covariance with age of child, age of mother, and social class as covariates.
Social versus biological parenting 151
TABLE 8.3 Means, Standard Deviations (SD), and F Values for Comparisons of Children’s Socioemotional Development between Family Type
Genetic relationship to
“A” scale Cognitive competence Physical competence Maternal acceptance Peer acceptance *p<0.05.
Donor Egg IVF Insemination Adoptive Donation Mother Father Interaction MeanSD Mean SD Mean SD Mean SD F F F 8.34.3 7.9 5.2 9.4 4.9 6.7 5.2 0.27 2.35 3.55 19.43.6 20.0 3.1 19.6 3.3 17.3 4.5 2.97 4.15* 1.28 19.33.2
19.7 2.8 19.4 3.4 19.6 2.9
0.11
0.00
0.40
17.33.5
17.0 3.1 16.5 3.2 17.3 2.6
0.61
0.61
0.17
18.34.0
18.5 3.9 19.2 3.4 17.0 3.6
0.17
1.05
0.67
There was no difference between groups for “A” scale scores, showing that the children did not differ with respect to the presence of emotional or behavioral problems. The only significant difference to emerge for the subscales of the Pictorial Scale of Perceived Competence and Social Acceptance was for cognitive competence [F(1, 124) =4.15, p<0.05], showing that children who were genetically unrelated to their father perceived themselves to be more cognitively competent than children who shared a genetic link with their father whether or not they were genetically related to their mother. Telling about Donor Insemination None of the parents with a child conceived by donor insemination, and only one set of egg donation parents, had told their child about their method of conception. A significant difference between groups was found with respect to parents’ attitude toward telling the child [χ2(3, 61)=16.37, p<0.001]. The donor insemination parents were most against telling, with 82% having decided never to tell compared with 38% of the egg donation parents. Although all but one of the children conceived using a donated gamete had not been told about their genetic origins, 51 % of donor insemination parents and 72% of egg donation parents had told maternal grandparents, and 20% of donor insemination parents and 63% of egg donation parents had told paternal grandparents. Although there was no difference between groups in the proportion of parents who had told maternal grandparents, there was a significant group difference with respect to telling paternal grandparents [χ2(1, 56)=10.28, p<0.01], such that fewer donor insemination parents than egg donation parents had told paternal grandparents. A significant difference between
Annual progress in child psychiatry and child development 2000–2001
152
groups was also found with respect to telling family friends [χ2(1, 61)=14.60, p< 0.01]. Only 30% of donor insemination parents had told friends, compared with 71% of parents with a child conceived using a donated egg. To examine whether the more secretive parents showed more negative outcomes with respect to quality of parenting or the socioemotional well being of the child, comparisons were carried out between those who had told a friend or family member that their child had been conceived by gamete donation and those who had not. No differences were found for any of the measures. Parents who planned never to tell their child or who were undecided on this issue were asked to give their reasons. For 49% of donor insemination parents and 69% of egg donation parents secrecy was attributed to a wish to protect the child. A wish to protect the mother was given as a reason by 19% of donor insemination parents and 23% of egg donation parents, and a wish to protect the father was given by 69% of donor insemination parents and 23% of egg donation parents. There was no group difference in the proportion of parents who reported the wish to protect the child or the wish to protect the mother as a reason for not telling the child. However, a significant difference between groups was found with respect to the wish to protect the father [χ2(1, 60)= 8.50, p<0.01], with a greater proportion of donor insemination parents than egg donation parents giving this as a reason for not telling the child. Interestingly, 85% of donor insemination parents and 69% of egg donation parents felt that there was no need to tell the child. There was no significant difference between groups for this variable. Of the three egg donation families for whom the donor was a family friend, one set of parents did not plan to tell the child and two sets of parents were undecided at the time of study. All three sets of parents believed that there was no need to tell, and gave the wish to protect the child as their only reason for maintaining secrecy.
DISCUSSION A number of differences between parents were identified according to their genetic relationship with their child. In families where the child was not genetically related to the mother, irrespective of the presence or absence of a genetic link with the father, mothers reported higher levels of marital satisfaction, and fathers reported lower levels of trait anxiety and less stress associated with parenting as measured by the total score of the PSI/SF and all three subscales. In families where the child was genetically unrelated to the mother but had a genetic link with the father, mothers reported lower levels of parental distress and more coordination with the father over discipline of the child. It seems, therefore, that in families where the child lacked a genetic relationship with the mother there were more positive findings regarding the emotional well being of parents, although no differences were identified in relation to quality of parenting. In families where the child lacked a genetic relationship with the father, the only difference to emerge for mothers reflected a more negative outcome in that they expressed less warmth toward the child whether or not they themselves were genetically linked to the child. No differences were identified with respect to the emotional well being or parenting quality of fathers. Thus, it appears that where differences exist between families according to the
Social versus biological parenting 153 presence or absence of genetic ties between children and their parents, the lack of a genetic link between the child and mother is associated with greater psychological well being among the parents but not with their quality of parenting. The lack of a genetic link between the child and the father does not seem to be related to the well being of parents or their quality of parenting apart from lower levels of mothers’ warmth to the child. An examination of interaction effects showed that in families where the child was not genetically related to the mother but was genetically related to the father, mothers reported less stress associated with parenting as measured by the total score and the parental distress subscale of the PSI/SF, and greater coordination with the father over discipline of the child. However, disputes between mothers and their genetically unrelated children were found to be more severe when there was a genetic link between the father and child. In terms of the socioemotional development of the children themselves, the only difference identified was that children who were not genetically related to their father perceived themselves as more cognitively competent whether or not they were genetically related to their mother. As a high proportion of sperm donors are students (Cook & Golombok, 1995), and many are medical students, this finding may reflect a real difference in cognitive competence between children who are, and those who are not, genetically related to their social father. It is perhaps surprising that families characterized by the absence of a genetic link between the mother and child generally showed more positive outcomes than families where the mother is genetically related to the child. A possible explanation for this finding is that raising a child who is not genetically related to the mother is perceived to be a greater undertaking for infertile couples than having a child through assisted reproduction, using the parents’ own gametes, or by donor insemination, where it is only the father who lacks a genetic link with the child. As a result, infertile couples who choose to raise a child who is genetically unrelated to the mother may be even more committed to parenthood, and consequently find parenting a more satisfying experience, than those who become mothers and fathers through other routes. The failure to find the predicted negative effect for families where the father lacked a genetic link with the child may stem from the same phenomenon; a strong desire to have children may outweigh any negative effects arising from the missing genetic link between the father and child. It is interesting to examine the findings of the present investigation alongside investigations of other family forms in which children are genetically unrelated to one or both parents. Studies of adopted children (Brodzinsky, Lang, & Smith, 1995), and children in stepfamilies (Hetherington, 1993; Hetherington & Clingempeel, 1992), have shown that the lack of genetic relatedness between a child and one or both parents can be associated with alienation and hostility between the parents and the child, and that the children are more likely to show psychological problems than children raised by their natural parents. However, the adjustment of children in such families is associated with a number of factors, including the age of the child at the time of the family transition (Brodzinsky & Schechter, 1990; Hetherington, Bridges, & Insabella, 1998). It appears, for example, that the earlier children enter into their adoptive family or stepfamily, the less likely they are to develop emotional or behavioral problems. The positive outcomes found in the present study for children raised from birth by a nongenetic parent (or from
Annual progress in child psychiatry and child development 2000–2001
154
early infancy in the case of the adopted children) suggests that the absence of a genetic relationship, in itself, does not lead to difficulties for parents or children. In considering the discrepancy between the findings of this study and the literature on adoption and stepparenting, it may be relevant that children born through egg or sperm donation do not experience the loss of an existing parent. Nor do they need to form relationships with new family members. There are a number of methodological limitations to the study, which need to be considered. First, only 62% of the donor insemination families participated in the research. Although this response rate is less than ideal, it is important to point out that these families are extremely concerned that by participating in research their secret will be exposed. From telephone conversations with parents of children conceived by donor insemination, it appeared that fear of disclosure was their primary reason for declining to take part. Given that none of the parents had told their child that he or she had been conceived by donor insemination, it was encouraging that almost two-thirds of these mothers agreed to be interviewed. A further potential source of bias in the study is that the egg donation families were recruited later than the other family types. This was unavoidable as the oldest egg donation children were still toddlers when data collection for the other families began. However, the interviewers had been involved in the earlier investigation and were thus fully familiar with the study techniques, and attitudes towards gamete donation had not changed markedly in the intervening years. In view of the difference in the ages of the children in the various family types, Pearson productmoment correlations were conducted to examine associations between children’s age and the parenting variables for which group differences were identified. None was significant, suggesting that the differences between groups did not result from differences in the age of the child. It may be the case that parents of genetically unrelated children are more likely to try to present their relationship with their child in the best possible light due to the negative attitudes that are sometimes associated with nongenetic parenting. Whereas such a social desirability effect cannot be ruled out, it was the parents who were genetically related to their child who obtained the highest scores on the “defensive responding” subscale of the PSI/SF, indicating that socially desirable responding was higher among genetic than nongenetic parents. A further difficulty concerns the reliance on maternal reports of father-child interaction. Nevertheless, other studies using father-mother pairs have found high agreement between father’s reports of their involvement with their child and mother’s assessments (Pleck, 1997). It is noteworthy that information about genetic parent-age had been kept secret from all of the children conceived by donor insemination and all but one of the children conceived by egg donation. In spite of the secrecy surrounding the method of their conception, it seems, from the lack of interaction effects, that these children were no more likely than adopted children (all of whom had been told about their genetic origins) and IVF children (all of whom were genetically related to both parents) to show emotional or behavioral problems, or a poor perception of their competence or social acceptance. As the use of donated gametes in the treatment of infertility has increased in recent years, so has the pressure on parents to disclose information about genetic origins to their child (Daniels & Lewis, 1996). With respect to adoption, a recent study has shown that,
Social versus biological parenting 155 contrary to the concerns of the critics of openness, providing information about birth parents did not confuse children or lower their self-esteem (Wrobel, Ayers-Lopez, Grotevant, McRoy, & Friedrick, 1996). However, the opinion of social policymakers that openness is beneficial for children contrasts sharply with the views of the parents in the present study who preferred not to tell. In this context, it is perhaps worth noting that contemporary family therapists are moving away from the notion that openness is good and secrecy bad, to the position that “it depends” (Chasin, 1993). In evaluating whether secrets are having a damaging effect on relationships, Papp (1993) argued that questions such as the effect of the secret on the functioning of family members, the effect on the communication between family members, and the effect of the unaware person finding out by accident or through someone else, should all be addressed. Whereas secrecy did not appear to have a negative effect on families with children aged up to 8 years old in the present investigation, it remains to be seen whether secrecy leads to difficulties as these children grow up. Interestingly, the parents who were most committed to secrecy were from families where the father rather than the mother lacked a genetic relationship with the child. The finding that one-half of the donor insemination parents and almost three-fourths of the egg donation parents had told someone other than their child about the use of a donated gamete in the child’s conception means that for these families there will always be a potential for disclosure from someone other than the parents themselves.
ACKNOWLEDGMENTS We would like to thank the Medical Research Council for funding this research. We are also grateful to Brian Lieberman and Marinos Tsirigotis for their help at the early stages of the study.
REFERENCES Abidin, R. (1990). Parenting Stress Index Test manual. Charlottesville, VA: Pediatric Psychology Press. Baumrind, D. (1989). Rearing competent children. In W.Damon (Ed.), Child development today and tomorrow (pp. 349–378). San Francisco: Jossey-Bass. Beck, A., & Steer, R. (1987). The Beck Depression Inventory manual. San Antonio, CA: Psychological Corporation. Bowlby, J. (1969). Attachment and loss. Vol. 1. Attachment. London: Hogarth Press. Bowlby, J. (1973). Attachment and loss. Vol. 2. Separation. London: Hogarth Press. Brodzinsky, D., Lang, R., & Smith, D.W. (1995). Parenting adopted children. In M. H.Bornstein (Ed.) Handbook of parenting, Vol. 3: Status and social conditions of parenting. Hove, U.K.: Erlbaum. Brodzinsky, D., & Schechter, M.D. (1990). The psychology of adoption. Oxford, U.K.: Oxford University Press. Burns, L.H. (1987). Infertility as boundary ambiguity: One theoretical perspective. Family Process, 26, 359–372.
Annual progress in child psychiatry and child development 2000–2001
156
Burns, L.H. (1990). An exploratory study of perceptions of parenting after infertility. Family Systems Medicine, 8, 177–189. Chasin, R. (1993). Foreword. In E.Imber-Black (Ed.), Secrets in families and family therapy (pp. vii–ix). New York: Norton. Clamar, A. (1989). Psychological implications of the anonymous pregnancy. In J. Offerman-Zuckerberg (Ed.), Gender in transition: A new frontier (pp. 111–121). New York: Plenum Press. Cook, R., & Golombok, S. (1995). A survey of semen donors: Phase II—the view of the donors. Human Reproduction, 10, 951–959. Daniels, K., & Lewis, G.M. (1996). Openness of information in the use of donor gametes: Developments in New Zealand. Journal of Reproductive and Infant Psychology, 14, 57–68. Daniels, K., & Taylor, K. (1993). Secrecy and openness in donor insemination. Politics and Life Sciences, 12, 155–170. De Paulo, B.M. (1992). Nonverbal behaviour and self-presentation. Psychological Bulletin, 111, 203–243. Golombok, S., Cook, R., Bish, A., & Murray, C. (1995). Families created by the new reproductive technologies: Quality of parenting and social and emotional development of the children. Child Development, 66, 285–298. Grossmann, K.E., & Grossmann, K. (1991). Attachment quality as an organizer of emotional and behavioural responses in a longitudinal perspective. In C.M.Parkes, J.Stevenson-Hinde, & P.Marris (Eds.), Attachment across the life cycle (pp. 93–114). London: Routledge. Grossmann, K.E., Grossmann, K., Spangler, G., Suess, G., & Unzer, L. (1985). Maternal sensitivity in northern Germany. In I.Bretherton & E.Waters (Eds.), Growing points of attachment theory and research. Monographs of the Society for Research in Child Development, 50, 233–256. Harter, S., & Pike, R. (1984). The pictorial scale of perceived competence and social acceptance for young children. Child Development, 55, 1969–1982. Hetherington, E.M. (1993). An overview of the Virginia Longitudinal Study of Divorce and Remarriage with a focus on early adolescence. Journal of Family Psychology, 7, 518–530. Hetherington, E.M., Bridges, M., & Insabella, G.M. (1998). What matters? What does not? Five perspectives on the association between marital transitions and children’s adjustment. American Psychologist, 53, 167–184. Hetherington, E.M., & Clingempeel, W.G. (1992). Coping with marital transitions. Monographs of the Society for Research in Child Development, 57(2–3). Hoffman, L.W., Thornton, A., & Marris, J.D. (1978). The value of children to parents in the United States. Journal of Population Studies, 1, 93–131. Hoopes, J.L. (1990). Adoption and identity formation. In D.M. Brodzinsky & M.D. Schechter (Eds.), The psychology of adoption (pp. 144–166). Oxford, U.K.: Oxford University Press. Isabella, R., Belsky, J., & von Eye, A. (1989). Origins of infant-mother attachment: An examination of interactional synchrony during the infant’s first year. Developmental Psychology, 25, 12–21.
Social versus biological parenting 157 Karpel, M. (1980). Family secrets. Family Process, 19, 295–306. Klagsbrun, M., & Bowlby, J. (1976). Responses to separation from parents; A clinical test for young children. British Journal ofProjective Psychology, 21, 7–21. Lamb, M.E. (1997). The role of the father in child development. New York: Wiley. Lutjen, P., Trounson, A., Leeton, J., Findlay, J., Wood, C., & Renou, P. (1984). The establishment and maintenance of pregnancy using in vitro fertilization and embryo donation in a patient with primary ovarian failure. Nature, 307, 174. Mahlstedt, P., & Greenfield, D. (1989). Assisted reproductive technology with donor gametes: The need for patient preparation. Fertility and Sterility, 52, 908–914. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood and adulthood: A move to the level of representation. In I.Bretherton & E.Waters (Eds.), Growing points in attachment theory and research. Monographs of the Society for Research in Child Development, 50, No. 209 (1–2), 66–104. Papp, P. (1993). The worm in the bud: Secrets between parents and children. In E. ImberBlack (Ed.), Secrets in families and family therapy (pp. 66–85). New York: Norton. Parke, R. (1995). Fathers and families. In M.Bornstein (Ed.), Handbook of parenting, Vol. 3 (pp. 27–63). Hove, U.K.: Erlbaum. Pleck, J.H. (1997). Paternal involvement: Levels, sources and consequences. In M.E. Lamb (Ed.), The role of the father in child development (pp. 66–103). New York: Wiley. Quinton, D., & Rutter, M. (1988). Parenting breakdown: The making and breaking of intergenerational links. Avebury, U.K.: Gower Publishing. Rust, J., Bennun, I., Crowe, M., & Golombok, S. (1988). The handbook of the Golombok Rust Inventory of Marital State. Windsor, U.K.: NFER-Nelson. Rust, J., Bennun, I., & Golombok, S. (1990). The GRIMS: A psychometric instrument for the assessment of marital discord. Journal of Family Therapy, 12, 45–57. Rutter, M., Cox, A., Tupling, C., Berger, M., & Yule, W. (1975). Attainment and adjustment in two geographical areas: I. The prevalence of psychiatric disorder. British Journal of Psychiatry, 126, 493–509. Rutter, M., Tizard, J., & Whitmore, K. (1970). Education, health and behaviour. London: Longman. Sants, H.J. (1964). Genealogical bewilderment in children with substitute parents. British Journal of Medical Psychology, 37, 133–141. Schaffer, J., & Diamond, R. (1993). Infertility: Private pain and secret stigma. In E. Imber-Black (Ed.), Secrets in families and family therapy (pp. 106–120). New York: Norton. Schechter, M.D., & Bertocci, D. (1990). The meaning of the search. In D.M. Brodzinsky & M.D.Schechter (Eds.), The psychology of adoption (pp. 62–92). Oxford, U.K.: Oxford University Press. Singer, L., Brodzinsky, D., Ramsay, D., Steir, M., & Waters, E. (1985). Mother—infant attachment in adoptive families. Child Development, 56, 1543–1551. Snowden, R. (1990). The family and artificial reproduction. In D.Bromham (Ed.), Philosophical ethics in reproductive medicine (pp. 70–185). Manchester, U.K.: Manchester University Press. Snowden, R., Mitchell, G.D., & Snowden, E.M. (1983). Artificial reproduction: A social
Annual progress in child psychiatry and child development 2000–2001
158
investigation. London: George Allen & Unwin. Spielberger, C. (1983). The handbook of the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Steer, R., Beck, A., & Garrison, B. (1986). Applications of the Beck Depression Inventory. In N.Sartorius & T.Ban (Eds.), Assessment of depression (pp. 123–142). Berlin: Springer-Verlag. Triseliotis, J. (1973). In search of origins: The experiences of adopted people. London: Routledge & Kegan Paul. Trouson, A., Leeton, J., Besanka, M., Wood, C., & Conti, A. (1983). Pregnancy established in an infertile patient after transfer of a donated embryo fertilized in vitro. British Medical Journal, 286, 835–838. Woollet, A. (1991). Having children: Accounts of childless women and women with reproductive problems. In A.Phoenix & A.Woollet (Eds.), Motherhood: Meanings, practices and ideologies (pp. 47–65). London: Sage. Wrobel, G.M., Ayers-Lopez, S., Grotevant, H.D., McRoy, R.G., & Friedrick, M. (1996). Openness in adoption and the level of child participation. Child Development, 67, 2358–2374.
PART II: PARENTING
9 Parenting among Mothers with a Serious Mental Illness Daphna Oyserman, Carol T.Mowbray, Paula Allen Meares, and Kirsten B.Firminger
In the past few decades, deinstitutionalization and community-based rehabilitation and support programs have increased the likelihood that women with serious mental disorders will be parents and will raise their children. This review describes what is known about the parenting of these women, focusing on diagnosis, child characteristics, and other contextual effects.
People with severe mental illness share the universal aspirations to form intimate relationships and have children. —Caton, Cournos, Felix, and Wyatt (1998, p. 86)
INTRODUCTION Warmth, nurturance, and provision of developmentally appropriate and consistent supervision, structure, and autonomy are the universal hallmarks of good parenting, from infancy through adolescence. Yet, the specific challenges and focal issues of parenting are closely linked to the particulars of the child’s developmental phase, the sociocultural context within which parenting is embedded, and attributes of the parents, including their mental health. The current review focuses on parenting among North American mothers with a long-term, serious mental illness—unipolar or bipolar affective disorder or schizophrenia (Dennis, Buckner, Lipton, & Levine, 1991; Gonzales, Kelly, Mowbray, Hays, & Snowden, 1991). Understanding the parenting process among these mothers is important because a sizable number of mothers experience a serious mental illness either before or after childbirth, with risk of onset remaining elevated during the early years of parenting. Risk of serious mental illness (SMI) is separable from simple postpartum depression, which itself affects 5% to 8.8% of women who give birth (Richards, 1990). Context of parenting A revised version of a paper submitted to the Journal in June 1999.
Parenting among mothers with a serious mental illness 161 and maternal mental health are inextricably linked. Being a mother of young children (under 5 years of age) increases risk of serious psychiatric symptoms such as anxiety and depression (Romans-Clarkson, Walton, Herbison, & Mullen, 1988) and psychiatric disorders are more likely when mothers are stressed and caring for multiple, younger children (Puckering, 1989). To the extent that problematic parenting at earlier developmental phases increases risk of problems in later developmental phases, risk due to maternal SMI may be cumulative. Further, because SMI is by nature episodic, with episodes lasting up to 2 years, children of parents with SMI are likely to experience more than one episode of parental mental illness influencing more than one developmental phase (Hammen, 1997). Maternal SMI is clearly a risk factor for children, who are at increased risk of being placed in alternative settings such as foster care (Oyserman, Benbenishty, & Ben Rabi, 1992) and of exhibiting behavior problems (Ghodsian, Zajicek, & Wolkind, 1984). In their lifetime, 32% to 56% of children of parents with SMI (schizophrenia or affective disorder) will themselves have a DSM diagnosable disorder (Amminger et al., 1999; Rieder, 1973; Waters & Marchenko-Bouer, 1980), with frequency depending on the diagnostic method, parents’ diagnoses, and length of followup. Moreover, as summarized by Rutter, Silberg, O’Connor, and Simonoff (1999), the heritability estimate of bipolar disorder is 80%; for depression, it is 34% to 48%; and for schizophrenia, it is 75%. Because appropriate parenting is importantly linked to the developmental phase of the child, as well as the particular sociocultural context in which it occurs, the risk the child experiences due to maternal mental illness is likely contingent on the interplay among onset of illness or timing of episodes, child’s developmental phase, and the sociocultural context in which parenting takes place. The interaction among these factors may be complex. Mothers with SMI may have difficulty dealing with some but not all developmental phases in their children’s lives, specific maternal diagnosis may interact with children’s temperament and characteristics, and some diagnoses may be more riskinducing in some cultural contexts than others (Gelfand & Teti, 1990). Finally, ethnic and racial groups differ in their likelihood of diagnosis for depression and schizophrenia (Garb, 1997; Ruttenberg, Finello, & Cordeiro, 1997), an effect that is as yet poorly understood. Therefore, the current review examines parenting during each phase of childhood (infancy, preschool, primary school, adolescence) separately, attempts to look at both the influence of diagnosis and the developmental needs of the child, and explores what is known about the interplay between child temperament and characteristics and maternal diagnosis. Where possible, we examine what is known about parenting for mothers who vary in their race, ethinicity, and socioeconomic status, in an effort to contextualize our understanding of the processes by which risk unfolds. The review focuses on mothers with a mental illness, rather than parents with a mental illness, for several reasons. First, although relatively little research in this area is available, a recent literature review suggested that parenting by a mother with SMI is more likely than parenting by a father with SMI (Nicholson, Nason, Calabresi, & Yando, 1999). In their own research on the caseload of a large urban community mental health center, Nicholson et al. found twice as many women who were known to be mothers than men who were known to be fathers. In addition, even though they are at higher risk of lifetime onset of mental illness, women with SMI are likely to fare better over time than
Annual progress in child psychiatry and child development 2000–2001
162
men, but also experience high rates of pregnancy—planned or unplanned (Mowbray, Oyserman, Zemenchuk, & Ross, 1995). Further, women with SMI are more likely to marry than are their male counterparts (Mowbray, Oyserman, Lutz, & Purnell, 1996; Mowbray et al., 1995). Lastly, later onset of mental illnesses (other than depression) for women means that women are more likely to have the opportunity to become parents prior to the onset of SMI (Kessler, 1997). That said, when fathers are at home during childhood, the effect of their depression or other mental illness on their parenting may be similar to or different from the effects of mental illness for mothers (Jacob & Johnson, 1997).
MOTHERHOOD IN CONTEXT Through midlife, parenting issues are often intertwined with other stresses that themselves are likely to affect parenting adversely (Cohler, Stott, & Musick, 1996; Stott, Musick, Clark, & Cohler, 1983; Stott et al., 1984). A recent review of pregnancy and the postpartum period for North American mothers with SMI highlighted the stresses that both precede and co-occur with episodes of mental illness, and that continue after the episode resolves (Mowbray et al., 1995). Women with SMI have a greater number of children and begin their childbearing early, and these women are at heightened risk both of poverty and of raising their children as single parents. Marital and family strife and victimization are likely, adding to the difficulties that these mothers and their children experience (Belle, 1990; Downey & Coyne, 1990; Olson & Banyard, 1993). There is some evidence that parenting can have positive and motivating effects for mothers with SMI. These women have identified motherhood as a central force keeping them involved with treatment, a key outlet for expression of feelings of care and concern, and a valued, normative social role (Mowbray, Oyserman, & Ross, 1995; Oyserman, Bybee, Mowbray, & Khang, 2000; Perkins, 1992). In these and other qualitative studies, women with SMI articulated the significance of having children, as well as their struggles to maintain or obtain custody and to achieve normal lives for themselves and their children (Nicholson, Sweeney, & Geller, 1998a; Sands, 1995; Schwab, Clark, & Drake, 1991). It seems plausible that motherhood would be a positive and central part of women’s lives, even if they have a serious mental illness. Nurturance and generativity (Erikson, 1968), as exemplified in parenting, are arguably the central tasks of adulthood. Parenting is a major social role, a normative sign of adult status, and an important developmental task (Belle, 1982; Cohler & Musick, 1983; Gizynski, 1985; Luster & Okagaki, 1993). Given its normative, social, and developmental status, parenting is likely to be central to one’s sense of self. Successes in this domain will therefore provide the self with a sense of worth and competence, whereas setbacks may be particularly stress-inducing, providing the basis for a variety of negative self-images (Markus & Cross, 1990; Stott et al., 1984). However, mothers’ wishes to parent well may be thwarted by difficulties due to specific aspects of mental illness (a main effect), the interplay between the child’s needs and the limitations on the mother’s capacities to meet these needs due to her mental
Parenting among mothers with a serious mental illness 163 illness, or the interplay between other contextual and clinical features (interaction effects). Therefore, it is critical that evidence for problematic parenting be assessed in terms of the mother’s clinical status and context as well as the age and other characteristics of the children. The current chapter attempts to do just that.
REVIEW OF PARENTING STUDIES Overview of Method To reduce variability in diagnostic technique and criteria, and to insure that results described women with access to community-based care, this chapter is confined to English-language research published since 1980. Studies were obtained via searches of both Medline and Psych Abstracts online from January 1, 1980, to January 1, 1999, using the following key words: maternal mental illness, parenting, child rearing, depression, mother-child relations, schizophrenia, bipolar, preschool/school-age children, adolescence, psychiatric symptoms, affective disorders, infants, mental or psychological disorders. To organize this review, we focused first on research documenting differences in parenting due to diagnostic differences—main effects of diagnosis—and then on research describing contextual effects. The vast majority of this work addresses parenting and SMI as defined, perceived, and reacted to within an American context; the few non-U.S. studies included here are explicitly noted. Every attempt was made to include all studies describing women with SMI. Studies using a research protocol such as the Beck Depression Inventory to assess symptoms of depression in an otherwise normal community sample were excluded because these are not typically measures of SMI. Finally, studies of women with postpartum depression only also were excluded, again because it is unclear if they meet SMI criteria. However, these criteria excluded all studies on the interplay between child characteristics and maternal SMI. Therefore, the section dealing with child characteristics includes research in which the child, but not the mother, was previously diagnosed, and maternal depression was assessed for research purposes. As a rule, studies are small in size and did not attempt to obtain random or representative samples. However, as will be seen, results form a relatively coherent picture, so that, despite limitations, the studies summarized do provide working hypotheses as to the effects of maternal SMI on parenting and the mother-child relationship (Lundy, Field, McBride, Abrams & Carraway, 1997). Analyzing this literature, we found that, although a variety of diagnoses are contrasted, research has typically compared women with depression to control mothers or to mothers with a single other diagnosis. For purposes of clarity, studies of mental illness generally, studies comparing diagnoses, studies taking into account other clinical characteristics, and studies describing contextual influences will be reviewed here separately, as will those on child characteristics, interpersonal factors, socioeconomic influences, and parental attitudes and beliefs. Within each section, studies are organized by age of child, because age has important consequences for the tasks of parenting. This holds true, as
Annual progress in child psychiatry and child development 2000–2001
164
well, for Tables 9.1a and 9.1b, which present maternal sample size and demographic data, including socioeconomic status (SES) and race/ethnicity (where available); maternal disagnostic categories and type of diagnostic assessment; and parenting measures used in each study. Review articles that did not present original data are discussed in the text but not presented in the tables. Similarly, studies focused on mothers with depressed symptoms that help illuminate the research on mothers with SMI are cited in the text but not the tables. Where child ages were diverse, the average reported age was used to locate a study in the table.
TABLE 9.1a Summary of Studies: Main Effects of Diagnosis
Maternal Sample Material Diagnosis or Parenting Dependent Size and Symptoms and Source Variables Study Demographics Diagnosed versus Control Mothers Infant (M=0–3 yrs) Gross 20 mentally ill, 20 Mental functioning not Maternal (1983) control formally assessed sensitivity/effectiveness in mother-child interactions Selfer et al. 262 families: Current and Past Mother-infant (1992) analyses based on Psychopathology Scales, interactions during 20 schizophrenia, DSM-II diagnosis distant, close, caretake, 38 depression,29 feed situations personality disorder, 36 controls; heterogeneous, matched for race/SES Preschool 24 treatment, 18 Previous diagnosis of Maternal Attitude Scale (M=3–5 controls; mixed psychosis mother child yrs) ethnicity, SES interactions Cohler, & Musick (1983); Musick et al. (1984) Cohler et 47 mentally ill, 48 Schizophrenia, Maternal and child al. (1980) control; middle schizoaffective, manic- behavior ratings class depressive, depression; official diagnosis Klehr et al. 34 mentally ill, 34 Schizophrenia, character Mother-child interaction (1983) control; 50% disorder, during feeding,
Parenting among mothers with a serious mental illness 165
Caucasian, 50% schizoaffective/affective structured task, married, high SES illness; unstructured play hospitalized/official situation diagnosis Stott et al. 65 mentally ill, 36 Schizophrenic, affective Mother-child interactions (1983) control; matched illness, character disorder, during feeding, Stott et al. for mother’s age formal diagnosis of structured task, (1984) & marital status, psychosis unstructured play race, SES School-age (M=5–12 yrs) & Adolescent (M=13–18 yrs) Scherer et al. 28 mentally ill, 30 Schizophrenia, (1996) control; majority schizoaffective, bipolar African-American, w/psychotic features, low SES, 50% official diagnosis/ single-parent hospitalization Affective Disorder Infant (M= 0–3 yrs) Cooper et al. 49 depressed, 49 Depression GHQ, PSE (1988) control; mostly Depression CES-D Stein et al. married, middle (1991) class Cohn 13 depressed; low&Tronick SES, 7 single (1989)
Free play, drawing exercise, puzzleconstruction, and clean-up time, Parent Perception Inventory
Mother-child interaction in structured play, strange situation Naturalistic observation, facetoface, structured social interactions Strange Situation, Block Q-Sort, home observation
Davenport et 7 manicManic depressive al. (1984) depressive parents, symptoms; SADS 20 control; matched on social class, race, education, family composition Field (1984) 12 depressed, 12 Depression; BDI, STAI Infants’ ability to control; low-SES detect mothers’ affective quality of displays and behavior in face-to-face play Field et al. 7 depressed, 9 Depression; BDI Sharing and
Annual progress in child psychiatry and child development 2000–2001
(1989) Field et al. (1990)
control AfricanAmerican, lower SES 24 depressed, 24 control AfricanAmerican, low SES
166
synchrony of behavior states and heart rate Depression; BDI
Attentive/affective behavior-state matching and synchrony in interactions Gaensbauer 7 bipolar, 7 control Bipolar disorder; SADS Free play observations, Strange et al. (1984) 4 mothers, 3 Situation fathers matched for age, race, SES Ruttenberg et 27 depressed, 19 Depression; CES-D Mother-infant al. (1997) control; Hispanic interaction, feeding with premature, very low-birth weight infant Preschool (M=3–5 yrs) Cox et al. 49 depressed, 27 Depressive symptoms; Mother-child social PSE interaction (1987) control; South London urban working class Kochanska et 54 mother-child Manic Maternal success, al. (1987) dyads, 33 control; depression/depression; compromise, use of middle-class SADS-L, GAS power, confrontation avoidance in mother-child interactions Radke14 bipolar, 42 Unipolar, bipolar, and Attachment, modified Yarrow et al. unipolar, 12m inor minor depression; SADS Strange Situation (1985) depression, 31 control; mostly white Radke49 unipolar, 24 Unipolar/bipolar; SADS- Mother-child Yarrow et al. bipolar, 45 control L observations of (1993) middle, upper predominant affect middle class, white Preschool (M=3–5 yrs) Cox et al. 49 depressed, 27 Depressive (1987) control; South London symptoms;PSE urban working class Kochanska 54 mother-child dyads, Manic
Mother-child social interaction Maternal success,
Parenting among mothers with a serious mental illness 167
et al. (1987) 33 control; middle-class depression/depression in mother-child SADS-L, GAS interactions RadkeYarrow et al. (1985) RadkeYarrow et al. (1993)
compromise, use of power, confrontation avoidance Attachment, modified Strange Situation Mother-child observations of predominant affect
14 bipolar, 42 unipolar, Unipolar, bipolar, and 12 minor depression, 31 minor depression; control; most white SADS 49 unipolar, 24 bipolar, Unipolar/bipolar; 45 control; SADS-L middle/uppermiddle class, white School-age (M=5–12 yrs) Hamilton et 16 unipolar, 13 bipolar, Unipolar depression, Mother-child al. (1993) 24 non-depressed, 11 bipolar disorder; SADS- conflict resolution chronic physical illness; L, RDC task videotaped primarily unmarried, Observations upper-middle SES, 30% nowhite Adolescent (M=13–18 yrs) Gordon 12 unipolar, 12 bipolar, Unipolar, bipolar Unstructured et al. (1989) 11 medically ill, 23 disorder; SADS, BDI, interaction task, control; mostly White, RDC mother-child middle/ upper-middle interactions, Peer class, unmarried Interaction Rating System Unipolar vs. Bipolar Depression School-age (M=6–12 yrs) Inoff10 bipolar, 27 unipolar, Unipolar, bipolar Family System Germain et 4 minor depression, 1 8 disorder; SADS-L, GAFTest, videotaped al. (1997) control families; mostly family interaction, white, middle class structured/ semistructured interviews, questionnaires, family meal, parentchild interaction tasks Tarullo et 34 unipolar, 16 bipolar, Unipolar depression, Observations of al. (1995) 27 nondepressed; bipolar disorder, SADS- mother-child primarily middle, class, L interaction, white differential treatment; maternal report of child problems
Annual progress in child psychiatry and child development 2000–2001
168
Depression versus Schizophrenia 88 diagnosed, 104 Infant Nonendogenous Mother-infant (M=0–3 yrs) control; subsample of psychosis, affective interactions during McNeil et McNeit et al. (1983): 46 illness, schizophrenia, feeding and al. (1983) diagnosed, 80 control; cycloid psychosis; RDC unstructured play Naeslund et prospective study, situation al. (1984) southwest Sweden Naeslund et al. (1985) McNeil et al. (1985); PerssonBlennow et al. (1986)
PerssonBlennow, Binett et al. (1988); PerssonBlennow, McNeil et al. (1988) Preschool (M=3–5 yrs) Goodman and 53 schizophrenic, 25 Schizophrenia, Home Scale Brumley depressed, 23 control; depression; DSM III (1990) mostly African-American, low income, single parent Sameroff et 29 schizophrenic, 58 Schizophrenia, Home and al. (1983) depressed, 40 personality depression and laboratory disorder, 57 control; 63% personality disorder; observations of Caucasian, 34% psychiatric mother-child AfricanAmerican, 3% interview interactions; Puerto Rican; 74% Schaefer and Bell married, mixed SES Parental Attitude Research Instrument School-age (M=6–12 yrs) Positive Rogosch et al. 48 mothers; low SES, Schizophrenia, involvement, (1992) minority sample schizoaffective, affective disorders; negative control, lax discipline; DSM-III clinical interview diagnosis social support
Parenting among mothers with a serious mental illness 169
Note. Studies are listed only once; for studies with results pertaining to multiple sections, listing is in the first relevant section. Studies using the same sample are cited together where possible; those with multiple age groups of children are cited by primary focus where possible. Abbreviations of measures follow, in alphabetical order: BDI—Beck Depression Inventory; CES-D—Center for Epidemiological Studies-Depression Scale; DSM-III—Diagnostic and Statistical Manual of Mental Disorders; FACES-III—Family Cohesion Scale of the Family Adaptability and Cohesion Evaluation Scales-III; GAF—Global Assessment of Functioning Scale; GAS—Global Assessment Scale; GHQ General Health Questionnaire; MMPI—Minnesota Multiphasic Personality Inventory: NIMH DIS—National Institute of Mental Health Diagnostic Interview Schedule; OPS— O’Leary-Porter Scale; PRS—Parental Role Stress; PSE—Present State Examination; PSI—Parenting Stress Index; RDC—Research Diagnostic Criteria; SADS—Schedule for Affective Disorders and Schizophrenia; STAI—State-Trait Anxiety Inventory. For more detail on measures, see cited study. TABLE 9.1b Summary of Studies: Interaction Effects
Maternal Sample Size Material Diagnosis or Parenting and Demographics Symptoms and Dependent Study Source Variables Community Functioning, Symptom Severity, Chronicity Infant (M=0–3 yrs) Campbell et 61 depressed, 60 control; Depression; SADS, Mother-child al. (1995) married, white and toy RDC interactions during play feeding, face-toface, Teti et al. 61 depressed; 43 control; Depression; BDI Feeding and free (1995) mostly white, married play and Strange Situation observations; PSI Preschool (M=3–5 yrs) Frankel & 30 depressed, 32 control; Depression; SADS, Early Relational Harmon mostly married, white RDC, BDI Assessment Scales (1996) and Strange Situation School-age (M=6–12 yrs) Harnish et al. 376 mothers and Depression; CES-D Mother-child (1995) children; national crossinteraction tasks sectional sample assessed w/Interaction Rating Scale Nolen40 depressed, 40 control; Depression; RDC, Joint puzzle task
Annual progress in child psychiatry and child development 2000–2001
Hoeksema et al. (1995) Oyserman, Bybee, Mowbray, & Khang (2000); Oyserman, Bybee, Mowbray, & McFarlane (2000)
170
mostly white, middleSADS, CEDS class, married 379 mentally ill: 43% SchizophreniaParental Locus of schizoaffective, Control Scale, PSI, depressed, 27% schizophreniaschizodepression, bipolar, Block Child affective, 19% anxiety or anxiety or adjustment Rearing Practices, adjustment disorder, 11% disorder; research Parenting bipolar, primarily DIS diagnosis, Involvement in Education Scale, AfricanAmerican urban symptomatology, mothers community Sensitivity to functioning score Children, Parenting Satisfaction, meaning of motherhood factor scores
Adolescent (M=13–18 yrs) Andrews et 59 mothers, 76 daughters al. (1990) British, working class
Focus on Context Infant (M=0–3 yrs) Gelfand et 71 depressed, 53 control; al. (1992) mostly white, married Zahn22 depressed, 22 control; Waxler et al. mostly white, married (1990)
Depression; PSE
Questionnaire regarding early adverse family experiences/ inadequate parenting
Depression; BDI
Mother-infant interactions during feeding and free play, PSI Depression; SADS- Mother-child L, BDI interaction in play and structured situation
Preschool (M=3–5 yrs) Free et al. 53 mentally ill, 31 control; Unipolar/bipolar; (1996) mostly white, SADS, GAS middle/uppermiddle class Goodman & 56 schizophrenic, 26 Johnson depressed, 29 control; (1986) mostly black Radke52 depressed, 38 control; Yarrow et al. mostly middle, upper(1994) middle class, white
Mother’s ability to interpret and communicate emotions Schizophrenia, Role Functioning depression, DSM III Scale
Unipolar/bipolar; SADS-L
Mother-child attachment relationship
Parenting among mothers with a serious mental illness 171
WebsterStratton & Hammon (1988)
situation and home observation 46 depressed, 49 control
School-age (M=6–12 yrs) Goodman et 20 unipolar, 19 control; al. (1994) white, middle class Hammen et 13 unipolar, 9 bipolar, 22 al. (1987) nondepressed, 14 medical; mostly white, middle class Hammen et 14 unipolar, 12 bipolar, 24 al. (1991); nondepressed, 14 medical; Hammen et mostly middle, upperal. (1990) middle class, white Hirshfeld et 30 mentally ill, 41 control; al. (1997) NIMH, DIS, SADS
Depression; BDI
Unipolar depression; BDI Unipolar, bipolar; SADS-L, BDI, shortform MMPI Unipolar depression, bipolar disorder; SADS-L
observed in experimental Home observation w/Dyadic ParentChild Interaction Coding Syst. Maternal expressed attitude Mother-child interaction behavior
Mother-child interactions, Peer Interaction Rating System Anxiety disorder; 5- Maternal expressed Minute-Speechmotion, Sample and lab observations of behavioral inhibition Hops et al. 27 depressed, 25 control; Depressive Family interaction (1987) matched on SES symptoms; SADS, observations of RDC, BDI nonverbal affective behavior Inoff15 bipolar, 22 unipolar, 24 Unipolar, bipolar; Mother-child Germain et control; mostly middle SADS-L, RDC interaction with two al. (1992) class, white age-adjacent children in free play and a snack Lundy et al. (1997)
20 depressed; lowSES, mixed racial composition Tarullo et al. 31 unipolar, 22 (1994) bipolar, 30 control; mostly white, married Adolescent (M=13–18 yrs) Davies & 443 adolescentWindle (1997) mother pairs; mothers primary caregivers; mostly
Depressive symptoms; CES-D Unipolar, bipolar; RDC, SADS-L
Observer ratings of 8 dimensions of mother-child interaction Mother-child interactions
Depression; CES-D
Family discord, adolescent girls’ adjustment, low family intimacy, parenting impairments w/FACESIII,
Annual progress in child psychiatry and child development 2000–2001
middleclass, married, white Tannenbaum 282 adolescents and Depressive & Forehand parents; white, symptoms; (1994) middle-class BDI
172
PRS, parental coping, and OPS Father-adolescent relationship as a buffer for maternal depressive mood measuring child’s internaling/externalizing problems
Parenting as the Link Infant (M=0–3 yrs) Jameson et al. 29 unipolar Unipolar Mother-infant interaction, free (1997) depression, 14 depression; play; Strange Situation task control; mostly SADS-L, RDC white, married School-age (M=6–12 yrs) Conrad & 14 unipolar Unipolar Mother-child interaction task depression, Hammen depression, 12 (1989) bipolar, 14 chronic bipolar medical disord., 24 disorder; no major psych, SADS-L, BDI, disorder, 80% white, shortform 12% Hispanic, 8% MMPI African-American Jacob & 41 maternal Major Mother-child, fathermotherJohnson depression, 50 depression; child, father-child, father(1997) paternal depression, SADS, RDC, mother interaction; problem50 control BDI, MMPI solving discussions Radke-Yarrow 39 unipolar, 24 Unipolar, Mother-child interaction; et al. (1995) bipolar, 32 control; bipolar; Strange Situation mostly middle/ SADS-L, upper-middle class, RDC, GAS white Teti & 48 depression, 38 Major PSI, Maternal Self-Efficacy Gelfand controls; mostly depression; Scale, mother-child BDI interaction, (1991) white, married feeding, free play Teti et al. 42 major depression, Major Maternal Self-Efficacy Scale, (1990) 10 dysthymia, 7 depression, mother-child interaction, adjustment disorder dysthymia, feeding, and free play w/depression; adjustment mostly white, disorder with married depression; BDI Note. See Table 1a footnote.
Parenting among mothers with a serious mental illness 173
SMI Mothers versus Control Mothers Nine of the studies reviewed compared control mothers to mothers with SMI, i.e., did not provide analyses by diagnostic category. These studies dealt primarily with parenting in the preschool years. When mentally ill and control mothers are contrasted without regard to diagnosis, the assumption is that mental illness per se, rather than the specific diagnostic category, impairs capacity for empathetic, nurturant, and appropriate parenting (Klehr, Cohler, & Musick, 1983). However, because the research reports in this section provide no analysis by diagnostic type, their findings cannot constitute evidence for or against this assumption. Further, although all of these studies reported that subjects and controls were matched on SES or race, only one of them provided this information about sample mothers and none of the studies took race or SES into account in their analyses; most participants appear to be white and middle to high SES. Infants. Two studies compared diagnosed and nondiagnosed mothers of infants. In a small clinical sample, Gross (1983) found no main effect of being mentally ill, but did find differences in infant outcomes related to parenting style (as described in the following, under context of parenting). In a second, larger study of parenting, with relatively few mothers who could be diagnosed as severely mentally ill, Seifer, Sameroff, Anagnostopolou, and Elias (1992) found that when infants were 4 months old and again when they were 12 months old, mothers with SMI were rated as less responsive to their infants when they were close, that is, within arm’s length of them. This finding presages findings of differences in mother-infant communicative synchrony described below in the literature on maternal depression. Preschoolers. Six studies conducted by Cohler and his colleagues focused on comparisons of diagnosed and nondiagnosed mothers of preschoolers. Control and diagnosed mothers differed particularly in the context of feeding. Mothers with a mental illness were less emotionally available, less reciprocal, less involved, less positive, and more likely to view denial as a means of dealing with childcare concerns (Cohler, Gallant, Grunebaum, Weiss, & Gamer, 1980; Cohler & Musick, 1983; Klehr et al., 1983; Musick, Stott, Spencer, Goldman, & Cohler, 1984; Stott et al., 1983, 1984). Compared to nondiagnosed mothers, mothers with SMI viewed establishing a reciprocal relationship with their young child as less important and differentiating their own needs from those of the child as less likely than did nondiagnosed mothers (Cohler et al., 1980). Klehr and colleagues (1983) found differences between seriously mentally ill and control mothers on five of seven interpersonal dimensions measured, even after a yearlong intensive intervention. In their discussion, the authors suggested that women with SMI experience more difficulty in negotiating developmental tasks and are not able adequately to resolve issues necessary for good interpersonal relationships and happy lives. School-aged Children and Adolescents. Although participants in the infancy and preschool studies were mostly white and middle class, the single study with school-aged children involved primarily African-American, low-SES participants with psychotic disorder diagnoses. School-aged children of these mothers were more likely to have behavior problems as assessed by parent and teacher child behavioral reports. Diagnosed mothers were rated by outside observers as exhibiting less encouraging parenting
Annual progress in child psychiatry and child development 2000–2001
174
behavior than low-SES comparison mothers. In addition, child behavior problems, as well as child selfcompetence ratings, were significantly predicted by maternal encouraging or negative-discouraging parenting style, controlling for psychiatric symptoms (Scherer, Melloh, Buyck, Anderson, & Foster, 1996). Maternal Affective Disorders Maternal depression is the most common diagnosis in research on parents with mental illness. The 15 studies reviewed here focus on chronicity or severity of mental illness; studies of other contextual issues or parenting attitudes are reviewed later in this chapter. In this section, although two studies reported controlling for race and SES, only one (Ruttenberg et al., 1997) involved Hispanic lowincome mothers; two (Field, Healy, Goldstein, & Gutherz, 1990; Field, Healy, & Leblanc, 1989) specified inclusion of African-American, low-SES mothers; and three additional studies (Cohn & Tronick, 1989; Cox, Puckering, Pound, & Mills, 1987; Field, 1984) specified non-minority group, lower-SES mothers. All examined parenting with infants, except the Cox study, which was located in London, England. Thus, again, most of our empirical knowledge about the influence of maternal depression comes from studies of white, middle-class mothers. Only studies with infants utilized more diverse samples. Infants. In a review article, Gelfand and Teti (1990) summarized the ways maternal depression and concomitant lack of maternal responsiveness can interfere with the child’s attainment of developmental goals, beginning with problems in infant attachment and language. Frankel and Harmon (1996) found that depressed mothers from a community sample rated themselves more negatively as parents and felt less attached to their infants. Mothers with a depressive episode in the 19 months after childbirth were lower in maternal rapport and facilitation than were control mothers (Cooper, Campbell, Day, Kennerley, & Bond, 1988; Stein et al., 1991). An 18-month follow-up study, comparing primarily white, middle-class control mothers to similar mothers with a bipolar disorder (who were married to men with unipolar or substance abuse diagnoses) also found evidence of problems in parenting (Davenport, Zahn-Waxler, Adland, & Mayfield, 1984). Specifically, diagnosed mothers were disorganized, inconsistent, ineffective and tense, less happy, and less active in interactions with their infants than were control mothers. The one study that did not find an effect used a small community sample of lowincome, Hispanic mothers of low-birth weight infants, half of whom received inhome prevention services; the study assessed parenting and infants at three points in time within the first 12 months of the baby’s life. Thus, it is unclear if the difference is due to the Hispanic sample, the intervention, or the fact that the diagnosis was made after delivery for purposes of the research (Ruttenberg et al., 1997). Mothers who are depressed may provide a less than optimal environment for their infants to learn social communicative skills; for example, a study of mothers with depressive symptoms suggested that their infants may be slower at learning to match happy face to happy vocalization (Lundy, Field, Cigales, & Cuadra, 1997). Though this study was not focused on mothers with SMI, it fits a general line of research suggesting that mothers diagnosed with depression are less affectively and interactionally synchronized with their infants. Field and her colleagues have highlighted the process of
Parenting among mothers with a serious mental illness 175 interactional synchronization (Field, 1984, 1986; Field et al., 1989, 1990). Building on similar research by Tronick and his colleagues (Cohn, Matias, Tronick, Connell, & Lyons-Ruth, 1986; Cohn & Tronick, 1987, 1988, 1989; Tronick, 1989; Tronick & Gianino, 1986a, 1986b) with mothers who are not seriously mentally ill, Field’s research suggests that although synchronicity normally rises in infants’ third to ninth month of life, as mothers learn to attune their affect to the child, this is less likely to occur for depressed mothers. Mothers of infants aged 3 to 12 months were also found to engage less in positive affect-eliciting play situations and much more in negative affect-eliciting behavior, such as poking, when they had depressive symptoms. Similarly, in a small clinical study of mothers with postpartum depression, depressed mothers were found to have interactions of shorter duration characterized by more variable maternal onset of response to infants (Zlochower & Cohn, 1996). All of this research suggests that the nature of the mother-infant interaction may be influenced by maternal depression. These findings parallel research in Sweden, where depressed mothers were found to talk to their infants less and received less social contact from their infants during feeding than did other mothers (Persson-Blennow, Naeslund, McNeil, & Kaij, 1986). Only one study contrasted bipolar with control mothers, using a small sample (N=7), at ages 12, 15, and 18 months (Gaensbauer, Harmon, Cytryn, & McKnew, 1984). Whereas control infants were less likely to have insecure attachment over time, infants of mothers with a bipolar disorder were more likely to evidence insecure attachment style. Preschool. Insecure and avoidant attachment has been found to be relatively frequent in families with major affective disorders. Insecure attachment is particularly likely when mothers have been depressed for a larger portion of the child’s life, have had a more severe depressive episode, received more intensive treatment for depression, or are single parents (Radke-Yarrow, Cummings, Kuczynski, & Chapman, 1985). When their children were preschoolers, mothers with an affective disorder were found to express more negative affect in extended observations of simulated family interactions and were less likely to achieve compromise with their children in videotaped interactions (Kochanska, Kuczynski, Radke-Yarrow, & Welsh, 1987; Radke-Yarrow, Nottelmann, Belmont, & Welsh, 1993). Converging evidence of the effect of maternal depression on parenting in the preschool years comes from a 6-month prospective study of South London urban, working-class mothers of 2-year-old children (Cox, Puckering, Pound, & Mills, 1987; Mills, Puckering, Pound, & Cox, 1985; Pound, Cox, Puckering, & Mills, 1985). These mothers were less able to sustain social interactions, more likely to ignore and use a less positive tone, and more likely to disengage, give fewer explanations, and ask fewer questions than were comparison mothers. In general, depressed mothers and their children were found to engage in more physical play and affectionate touching but not to engage in sustained or “linked” verbal interchanges. It was not that depressed mothers were not warm and affectionate with their children; it was that their warmth, expressed in physical contact, was not in tune with their child’s current needs. School-aged Children. In a laboratory-based, conflict resolution task, using a racially heterogeneous sample, mothers with affective disorders were more negative in their affective style (critical, intrusive, guilt-inducing statements) than control mothers, but had
Annual progress in child psychiatry and child development 2000–2001
176
no more problems communicating in interactions with their 8 to 16-year-old children (M=12; Hamilton, Jones, & Hammen, 1993). Adolescence. Mother-child interactions were observed as part of a longitudinal study of mostly Caucasian, middle-class, high-risk, 8- to 16-year-olds (M=about 13 years) and their unipolar and bipolar diagnosed mothers. It was found that mothers with unipolar depression were less positive, more critical, and showed less task-oriented behavior in a conflict discussion task with their children (Gordon, Burge, Hammen, & Adrian, 1989). Unipolar versus Bipolar Depression Preschool to Adolescence. None of the comparisons of unipolar and bipolar disorders involve infants. Some studies reviewed in the previous section included a comparison of mothers with unipolar depression and those with a bipolar diagnosis (Gordon et al., 1989; Hamilton et al., 1993; Kochanska et al., 1987; Tarullo, DeMulder, Ronsaville, Brown, & Radke-Yarrow, 1995). Effects were more severe for mothers with unipolar as compared to bipolar disorder. (Additional studies, discussed in the following under child characteristics, show interactions of maternal unipolar vs. bipolar diagnosis with child characteristics.) A single study with school-aged children suggested higher risk for bipolar mothers; insecure and avoidant attachment was particularly frequent in children of mothers with a bipolar diagnosis; and, though interaction problems overall were evident when the mother had a unipolar depressive disorder, mothers with a bipolar disorder showed more anger in family interactions (InoffGermain, Nottelmann, & RadkeYarrow, 1997). Depressed, Schizophrenic, and Control Mothers Relatively few studies distinguish parenting of mothers diagnosed with schizophrenia from control mothers or those diagnosed with depression. Of the four studies with North American samples, two look at preschool and two at school-aged children; all involve samples with a mixed or low SES and racially or ethnically diverse samples. Infancy. In the past two decades, the only comparisons across these diagnoses with infants came from a large-scale, longitudinal study of women giving birth in southwestern Sweden over a 4.5-year period, in which children and their mothers were followed from 3 days postpartum until the child was 1 year of age. Mothers later hospitalized with a variety of nonorganic psychoses, including schizophrenia and affective disorders, were compared to never-hospitalized mothers (Malmquist-Larsson, 1984; McNeil et al., 1983; McNeil, Naeslund, Persson-Blennow, & Kaij, 1985; Naeslund, Persson-Blennow, McNeil, & Kaij, 1985; Persson-Blennow, et al., 1986; PerssonBlennow, Binett, & McNeil, 1988; Persson-Blennow, McNeil, & Blennow, 1988). In this sample, fewer differences were found when mothers were observed feeding their children than when they were observed at play with their children, although laterhospitalized and control mothers differed in both feeding and play situations. Specifically, diagnosed mothers were more uncertain about their infant’s needs, provided less social contact, were less involved with their infant, and were less able to create a positive emotional climate. These differences were especially pronounced in the first 6 weeks postpartum,
Parenting among mothers with a serious mental illness 177 diminishing thereafter. Specific differences were found across diagnosis in the later-hospitalized group. Mothers diagnosed with cycloid psychosis (a variety of schizoaffective diagnosis) differed most from comparison mothers, followed by those diagnosed as having schizophrenia. At the 1-year assessment period, infants of these mothers were less likely to exhibit secure attachment, and more likely to exhibit deviant fear of strangers than were comparison infants. Mothers diagnosed as having an affective disorder were observed to have interactions that were at least as positive as the comparison group until the 1-year assessment. At this time, these mothers were found to talk less with their infant during feeding; reciprocally, the infant showed less social contact with the mother. Preschool. In a longitudinal study following an American sample of racially and economically diverse mothers from pregnancy through 4 years of age, differences in the group with schizophrenia occurred mostly in the first 4 months (Sameroff, Seifer, & Barocas, 1983). In these first months, mothers with schizophrenia were less involved; however, all other significant betweengroup differences focused on mothers with depression, who were more anxious and less socially competent. At 30 months, these mothers reported that their infant showed less adaptive behavior. Conversely, in terms of cognitive development, the children of mothers with schizophrenia showed highest risk. In a subsequent study comparing African-American control mothers to African-American mothers with schizophrenia or depression, Goodman and Brumley (1990) found that mothers with schizophrenia had the lowest quality parenting. School-aged Children. A small clinical study of hospitalized mothers with schizophrenia or an affective disorder failed to find significant differences on self-rated maternal sensitivity by diagnosis (Rogosch, Mowbray, & Bogat, 1992). In a large-scale study of mothers with SMI receiving treatment in a metropolitan area, Oyserman, Bybee, Mowbray, and Khang (2000) found that, although diagnoses did not differentiate maternal responses, level of symptomatology and community functioning did predict maternal identity. Positive maternal identity, in turn, related to more self-reported nurturant parenting styles.
DIAGNOSES VERSUS OTHER CLINICAL CHARACTERISTICS Although studies of maternal depression generally suggest that it is a clear risk factor for children (Hammen, 1997), its relative risk is difficult to determine for a number of reasons. First, as is shown in the present review (see Table 9.1a), research normally involves specific diagnostic comparisons between mothers with depression and one other group, making ranking of the relative risk of each diagnosis impossible. Second, research typically focuses on parenting within a specific child age group (e.g., 0 to 3 years) or does not analyze parenting by age of child. The bulk of current research contrasts depressed and control mothers (depression is studied in 14 of 23 studies with infants, 8 of 17 studies with preschoolers, 15 of 23 studies with school-age children, and all studies with adolescents). Third, the published studies we found did not analyze the possible interplay among maternal diagnosis, race, and SES. Further, when mentioned, race of participants was nearly always white, and SES nearly always middle-class or higher,
Annual progress in child psychiatry and child development 2000–2001
178
severely limiting generalizability of findings. Thus, when describing the influence of maternal mental illness on parenting among infants, our knowledge base comes primarily from research on the influence of depression in white and middle-class mothers. Few studies have included systematic comparisons across the spectrum of psychiatric diagnoses, and inclusion of racial and socioeconomic factors in analyses is almost nonexistent. In addition, as will be elaborated on in the next section, only seven of the 66 articles reviewed attempted to isolate the effect of diagnosis from that of other clinical variables such as chronicity, symptom levels, and current com-munity functioning. Further, because the seven studies that assessed these other clinical variables primarily involved maternal affective disorders, our knowledge of the effect on parenting of diagnosis, as compared to symptom severity or chronicity, is limited to depression. Community Functioning. In a sample of mothers with a unipolar or bipolar depression diagnosis and school-aged children, Inoff-Germain and colleagues (1997) found that community functioning related strongly to almost all the parenting variables assessed. Although the independent effect of functioning, controlling for the effect of diagnosis, was not presented, the authors showed that mothers who functioned better in the previous year had fewer affective problems in family interactions. In our own ongoing research with low-SES, African-American mothers with affective and schizophrenic disorders, we also found a promotive effect of current community functioning on parenting outcome variables, with no effect of diagnosis (Oyserman, Bybee, Mowbray, & McFarlane, 2000). Symptom severity. Four studies described the significant influence of severity of symptoms on outcomes. In a longitudinal study of mothers with depression, severity of depression was predictive of insecure attachment when the infant was both 1 and 2 years of age (Teti, Gelfand, Isabella, & Messinger, 1995). Women with “double depression” (unipolar episodes and dysthymia) were less emotionally available and more negative, and had pre-school aged children who were less securely attached than did women with simple depression (Frankel & Harmon, 1996). In a national cross-sectional sample of schoolaged children, severity of depressive symptoms was inversely related to the quality of the mother-child relationship (Harnish, Dodge, & Valente, 1995). A second large-scale study of mothers with depression used a pattern-matching test of solvable puzzles to show that level of mothers’ depressive symptoms correlated significantly with their negative affect and with their school-aged children’s enthusiasm, persistence, and frustration on the tasks overall (NolenHoeksema, Wolf son, Mumme, & Guskin, 1995). Chronicity. Six studies reported influence of psychiatric history on parenting in North American samples. Lifetime history (Gordon et al., 1989), number of episodes and duration of current episode (Nolen-Hoeksema et al., 1995), and portion of child’s lifetime during which mother experienced depressive episodes (Radke-Yarrow et al., 1993) did not influence parenting behavior in the three studies cited. However, three other studies did find a negative impact of chronicity of mental illness on parenting. Campbell, Conn, and Meyers (1995) found that chronicity of maternal depressive symptoms influenced motherchild interactions at 2, 4, and 6 months of age. Rogosch et al. (1992) reported that frequent recent rehospitalizations significantly predicted lower parenting sensitivity in mothers of school-aged children. Lastly, Gross (1983) found that chronicity, rather than mental illness diagnosis per se, negatively affected parental sensitivity and effectiveness.
Parenting among mothers with a serious mental illness 179 A single study with a British sample also suggests an impact of chronicity: Andrews, Brown, and Creasey (1990) found significant differences in adolescent or young adult daughters’ reports of early inadequate parenting based on whether mothers had a single short episode or a chronic, recurrent psychiatric disorder.
FOCUS ON CONTEXT Explicating the parenting context of mothers with SMI is critical because contextual factors may bolster or constrain mothers’ abilities to provide ageand culturally appropriate, “good enough” parenting. This section examines the literature on influences of children’s ages and other characteristics, support available to the mother, and ecological maternal stresses such as poverty. Child’s Age or Developmental Phase The child’s age or developmental phase can affect maternal behavior and responsivity in a number of ways. First, traditional American theories of child development hold that maternal illness episodes during the child’s first year of life and frequent separations may be quite damaging. This applies particularly to infants and younger children, because it is through interactions with their primary caregivers that their basic social skills, sense of self, and self-competence are established (Lundy, Field, Cigales, et al., 1997). As children develop, mature, and become more independent, their outcomes may be shaped less by parents and more by interactions with nonfamilial adults and peers, so that weaker effects with age result (Harris, 1995). For example, Klehr and colleagues (1983) suggested that early separations are more predictive of problems in the child’s capacity for object relations and use of defense mechanisms than are specific maternal diagnostic category or chronicity of illness. Alternatively, it is possible that maternal mental illness is particularly problematic after infancy. For example, if the effects of maternal psychiatric condition and parenting problems are cumulative, then we might expect stronger relationships between maternal disorder and problematic child outcomes for older children. Unfortunately, systematic comparisons of results across various developmental stages of childhood are rare. When tested, age effects are explored either by examining interactions between maternal psychiatric variables and children’s ages or by utilizing child age as a covariate. Several of the more comprehensive investigations utilized a narrow child age range (e.g., Nolen-Hoeksema et al., 1995), and studies with a wider child age range did not test for age differences (e.g., Hamilton et al., 1993). However, Inoff-Germain et al. (1997) did report that the highest rates of behavioral problems were for younger children (average age about 9) versus older children (average age about 13), especially in the children of mothers with a bipolar disorder. Perhaps because studying both child age and other relevant factors requires large and diverse samples, it is rarely undertaken. When possible age effects have been studied for children within a particular developmental phase, none was found in preschool (Free, Alechina, & Zahn-Waxler, 1996; Goodman & Brumley, 1990; Kochanska et al., 1987;
Annual progress in child psychiatry and child development 2000–2001
180
Radke-Yarrow et al., 1993; RadkeYarrow, Zahn-Waxler, Richardson, Susman, & Martinez, 1994) or school-age samples (Hirshfeld, Biederman, Brody, Faraone, & Rosenbaum, 1997). In addition, Tarullo et al. (1995) and Inoff-Germain et al. (1997), both having examined relationships between mother-child interactions and maternal psychiatric variables for younger versus older siblings, found no differential effects. However, maternal age, children’s age, and duration of mental illness are likely to be confounded; as mothers age, so do their children and there is a parallel increase in number of years with a mental illness diagnosis. To the extent that some problematic symptoms—such as active symptoms—are associated with the onset of a mental illness, then younger children are more likely to experience parenting from a severely impaired mother. This multicausality has yet to be addressed empirically. Child/Parent Characteristics In addition to developmental shifts, parenting and parenting contexts interact with child characteristics and behavior recursively and in multiple ways over time to affect child outcomes (Greenberg, 1998; Patterson, 1998; Reid, 1998; Sameroff et al., 1983). Two recent studies of mothers with depressive symptoms suggest that newborns of these mothers show less orientation toward faces, less interest in them, and more “pre-cry” responses (Lundy, Field, & Pickens, 1996) and that, at 4 months, these infants are rated as more vulnerable and engage in less exploratory play (Field et al., 1996), suggesting that parenting may be more difficult and less rewarding for depressed mothers of such more difficult infants. Child Gender. A few studies focus on the interplay between child gender and parenting. First, in a study of unipolar, bipolar, and control mothers (RadkeYarrow et al., 1993), unipolar depressed mothers of daughters had more sustained bouts of negative affect, more negative affect overall, and more anxious, sad, and downcast affect than other mothers. In interactions with sons, depressed mothers were no more negative than were control mothers. RadkeYarrow and colleagues (1993) found a significant correlation between the total amount of negative affect expressed by depressed mothers and their daughters. Similarly, Tarullo, DeMulder, Martinez, and Radke-Yarrow (1994) found that, among school-aged offspring of mothers with affective, especially bipolar disorder, children were less engaged in interactions with mothers and generally less happy and at ease with their mothers; mothers were more negative and critical, especially if children had behavioral problems. These findings are amplified for mother-daughter interactions and increase as daughters reach adolescence. This effect may be mediated by family functioning: Davies and Windle (1997) found that family discord mediates the effect of maternal depressive symptoms on adolescent girls’ adjustment. Overall, girls may be more responsive to maternal depression, such that lower levels of maternal depression are needed to elicit caring responses from girls than they are from boys (Radke-Yarrow et al., 1994). Child Behavior. Two comprehensive reports studying parenting of mothers with a mental illness addressed interactive effects of mother and child variables. Utilizing data from the UCLA Family Stress Project and structural equation modeling, Hammen, Burge, and Stansbury (1990) found that child outcomes are best predicted by the interaction
Parenting among mothers with a serious mental illness 181 between maternal functioning and child behavior. Second, Cox and colleagues (1987) reported that when the children of depressed mothers showed significant behavior disturbance, mothers were less likely to connect with or follow the cues of their children. Eight smaller, clinical studies corroborate these findings. Gelfand, Teti, and Fox (1992) found interaction between maternal depression and infant temperament, such that mothers of difficult infants reported increased parenting stress. In another study of mothers with depression, mothers’ parenting problems and childhood aggression at age 2 predicted child externalized problems at age 5 (Zahn-Waxler, Iannotti, Cummings, & Denham, 1990). School-aged children with externalizing disorders were found to have fewer interactions with and less positive affect toward their depressed mothers (Lundy et al., 1997). Mothers with an anxiety disorder were more critical of their behaviorally inhibited school-aged children (Hirshfeld et al., 1997). Mothers with a history of unipolar disorder made more self-blaming statements, but also were more likely to have schoolaged children with a history of depression or anxiety (Goodman, Adamson, Riniti, & Cole, 1994). In a series of hierarchical regression analyses, Gordon et al. (1989) showed that mothers’ chronic stress reduced the quality of maternal communication in a discussion-interaction task with adolescent children above and beyond the contribution of current symptoms. Children who viewed their mothers as having more symptoms were rated as less competent and rated their own self-worth as lower; this effect was most pronounced when mothers and their children had conflicting reports of maternal symptoms (Scherer et al., 1996). Finally, in a study of the interactions between mothers with depression and two close-in-age preschool or school-aged siblings, mothers with bipolar depression were found to have the most negatively interacting family groups, whereas mothers with unipolar depression were found to have a dearth of negativity compared with control families (Inoff-Germain, Nottelmann, & Radke-Yarrow, 1992). Social/Interpersonal Context Marital discord and social isolation are common for seriously mentally ill mothers and their children (Cox et al., 1987; Downey & Coyne, 1990), as are conflicts with extended family (Nicholson, Sweeney, & Geller, 1998b) and histories of sexual abuse (Alexander & Muenzenmaier, 1998). Moreover, once in place, marital conflict is likely to persist between episodes of depression (Puckering, 1989). Further, depressed persons tend to marry spouses with a psychiatric illness, a family history of psychopathology (Davenport et al., 1984; Merikangas & Spiker, 1982; Wang & Goldschmidt, 1994), or substance abuse (Davenport et al., 1984). When depressed persons have psychiatrically disturbed spouses, their own symptoms are more severe, and child symptoms (Zahn-Waxler et al., 1990) and marital and family disturbances are more likely (Puckering, 1989; Quinton, Rutter, & Liddle, 1984). Compared to control mothers, mothers diagnosed with depression are more likely to have adverse life experiences, particularly in the domain of close relationships (Cox et al., 1987; Downey & Coyne, 1990; Goodman & Johnson, 1986; Mills et al., 1985; Pound et al., 1985; Webster-Stratton & Hammond, 1988). For mothers with a depression diagnosis, marital discord contributes to negative interactions with children (Hops et al., 1987). Maternal depression, stressful life events, and child depression are likely to
Annual progress in child psychiatry and child development 2000–2001
182
interact, such that the experience of stressors might explain the temporal association of depressive episodes in mothers and diagnoses of depression in their children (Hammen, Burge, & Adrian, 1991; Hammen et al., 1987). In addition, multiple hospitalizations and the chronicity of mental health problems can contribute to marital breakups and reduce paternal involvement. At least one report has suggested that this is likely when husbands give up hope of their spouse’s eventual recovery and begin to fear that their children will become mentally ill like their mother (Cohler & Musick, 1983). Both unstable living arrangements and erratic contacts with fathers have been found to characterize families with a maternal schizophrenia diagnosis (Caton et al., 1998). Absent fathers are a risk factor, in that a positive fatheradolescent relationship can act as a significant buffer for maternal depressed mood (Tannebaum & Forehand, 1994), and better outcomes for mothers and their children are reported when the spouse or partner is supportive (Puckering, 1989). Higher levels of chronic stress, lower rates of positive life events, and single parenthood are all predictive of more negative maternal affective style (Hamilton et al., 1993). In a small sample of hospitalized mothers, social support correlated with increased self-esteem, which in turn correlated with more adaptive parenting style (Rogosch et al., 1992). Lack of social support, family discord, and marital stress have all been linked to negative outcomes for the children of mothers with SMI. However, in a small study of low-income African-American mothers, social support received in the past month did not improve parenting (Goodman & Johnson, 1986). Economic Context Stresses of single parenting under conditions of poverty with concomitant social isolation and marital discord can at least partly account for the initiation or maintenance of child disorders (Ghodsian et al., 1984; Hammen et al., 1987; Sameroff et al., 1983). Mothers with less education and lower SES, whether diagnosed with a mental illness or not, may be less effective in their maternal roles. Harnish et al. (1995) found a significant relationship between motherchild interaction quality and SES for a large representative sample of children (see also Gross, 1983). And women with SMI are at risk of chronic economic hardships (Belle, 1982; Dennis et al., 1991; Goodman & Johnson, 1986; Sameroff et al., 1983), such as poverty and overcrowding. These factors, as well as living on an upper story in a building with no elevator and spending many days alone with one’s baby, predicted child behavior problems in a longitudinal study of British-born mothers in central London, with little additional variance explained by maternal depression (Puckering, 1989). Other Contextual Variables Because separations from children may be particularly damaging, community-based treatment should be used to mediate the effects of stressors on mothers with SMI. For example, Free et al. (1996) reported that depressed mothers receiving treatment were better able to speak accurately with their preschool children about emotions than were depressed mothers not receiving psychotherapy. However, it is unclear to what extent women enrolled in ongoing mental health interventions typically receive any treatment
Parenting among mothers with a serious mental illness 183 that addresses, or even considers, their needs as mothers or the needs of their children. For example, 44% of psychiatric inpatient records at a private teaching hospital did not indicate whether the patient had children; when parenthood was documented, whereabouts of children were recorded in only 20% of cases, and children were contacted in only 32% of cases (DeChillo, Matorin, & Hallahan, 1987). In the United States, several outpatient programs supporting mothers with SMI have been reported (Cohler & Musick, 1983; Gonzales et al., 1991; Oates, 1988; Stott et al., 1984; Tableman, 1987; Waldo, Roath, Levine, & Freedman, 1987), but their numbers are limited (Oyserman, Mowbray, & Zemencuk, 1994).
PARENTING AS THE LINK BETWEEN PSYCHIATRIC VARIABLES AND CHILD OUTCOMES A number of authors have articulated possible frameworks through which psychiatric variables (particularly depression) affect parenting and thus contribute to problem behavior in children (Gelfand & Teti, 1990; Hammen, 1991b; Puckering, 1989). For example, Hammen (1991b) suggested that depression may reduce maternal receptivity to children’s cues. Lack of adequate and sensitive caregiving during infancy affects the infant’s later capacity to regulate tension and experience personal integrity. This dampened capacity, in turn, results in a lifelong deficit in the child’s ability to soothe tensions and manage transactions in the world, creating a sense of personal depletion and despondency (see Hammen, 1997, for a more detailed review of this literature). Some empirical evidence supports the notion that mothers with depression are at risk of more problematic parenting styles, increasing risk for their children. Among depressed mothers of infants in a clinical sample, those who felt efficacious as parents were more competent in their observed parenting interactions than were those who did not (Teti, Gelfand, & Pompa, 1990), and efficacy fully mediated the relationship between depression and parenting (Teti & Gelfand, 1991). Depressed mothers of 13- to 29-monthold toddlers were less able to focus on and sustain interactions with their toddler than were nondepressed mothers (Jameson et al., 1997). In a sample of depressed and nondepressed mothers of children, aged 3 to 8 years, who were receiving services for their behavior problems, depressed mothers reported more negative attitudes about their children (Goodman et al., 1994) and their own parenting, and more parental stress (Webster-Stratton & Hammond, 1988). Use of affectively charged, hostile, critical, and negative statements in describing one’s children has been linked to maternal history of depressive episodes (Goodman et al., 1994). Moreover, Goodman et al. (1994) found that positive parenting moderated outcomes for children of depressed, but not of well mothers. Depressed mothers of children 3 to 16 years of age expressed more dysphoric and less positive parenting affect; and positive parenting reduced the likelihood that their children would respond aggressively to the mother (Hops et al., 1987). Although maternal depression is a risk factor, parenting variables were found to be significant predictors of child social behavior and IQ, controlling for maternal psychiatric diagnosis (Goodman & Brumley, 1990). Independent of diagnosis, when mothers believed in encouraging individuation, mastery, and positive
Annual progress in child psychiatry and child development 2000–2001
184
interactions, their children performed better on cognitive and socioemotional measures (Cohler et al., 1980). Maternal anticipatory and respectful guidance for children 2 years of age predicted fewer problems for these children when they were 5 years old (ZahnWaxler et al., 1990). Mothers low in maternal effectiveness and sensitivity, whether mentally ill or not, had unrealistic expectations of their children’s cognitive capacities and were unable or unwilling to allow their children to take initiative in play, requiring that children control their own interests and affect and follow maternal interests. At the same time, children of these mothers were less able to control their anger and more resistant to maternal bids for affection than were children of more sensitive and effective mothers (Gross, 1983). Maternal depression increases risk of family conflict, which in turn increases adolescent psychiatric symptoms (Davies & Windle, 1997). Maternal depression may also reduce mothers’ positivity bias: Although mothers generally underestimate their child’s problems, depressed mothers are accurate in reporting the behavior problems of their school-aged children (Conrad & Hammen, 1989). The one study we found that contrasted the effects of maternal and paternal depressive symptoms concluded that maternal symptoms had a larger negative effect on overall family positivity and the positivity of parent-child interactions specifically, especially when the child was a daughter (Jacob & Johnson, 1997). However, even when children are able to form a secure attachment to their affectively disordered mother, the close nature of the bond may function to channel children toward affectively disordered coping styles later in life. Thus, in a longitudinal study of the interplay between security of attachment and maternal mental illness, children at age 6, especially girls, who had been securely attached as infants to mothers with severe depression were more likely to display depressive symptoms themselves, whereas children who were insecurely attached as infants to mothers with a bipolar disorder did not display problematic levels of anxiety when they were 6 years old (Radke-Yarrow et al., 1995).
CONCLUSIONS This review of what is known about parenting of women with a serious mental illness suggests that detrimental effects of maternal depression emerge by the time the infant is 1 year of age, and that a diagnosis of depression may be useful in making predictions about some aspects of parenting behavior and about mothers’ parenting style. Depressed mothers are less likely to develop the synchronized interactions with their infants that aid the child’s emerging sense of competence. They are more likely to have anxiously attached toddlers, even though they maintain extensive physical contact with their children. When their children are school-aged and adolescent, mothers with depression are at risk of continued negative interactional styles. They are also more prone to hold parenting attitudes that do not result in responsive and effective parenting. Results are similar when general, nondiagnosis-specific comparisons are made. When contrasted with bipolar diagnosis, unipolar depression appears more detrimental. Fewer studies of schizophrenia are available, and these deal mainly with infancy and the
Parenting among mothers with a serious mental illness 185 preschool years. Mothers with schizophrenia are reported to be less involved, more uncertain, and less able to create a positive climate for their infant in the first few months of life. During the school-age years, their children are most likely to have cognitive deficits. A few studies emphasized the importance of symptom severity, chronicity of mental illness, and level of community functioning in predicting both parenting behavior and child outcomes when mothers have depression. However, the literature makes it clear that mental illness is likely to be only a small part of the total risks mothers experience; these include family disruptions and conflicts, single-parent status, social isolation, and financial and other stresses associated with living in poverty. These difficult life circumstances are often concomitant with chronic long-term depression and potentiate and exacerbate risk. Further, very few studies focus on issues related to parenting of adolescents—a particularly glaring gap, because many mental health problems begin to emerge most sharply at this age. Although negative effects of maternal mental illness have been reported across the full spectrum from infancy through adolescence, little attention has been paid to longitudinal assessment of risk over time and developmental phases. Perhaps due to the crosssectional nature of most research, no evidence currently exists that particular phases are more risky than others. With regard to child characteristics, some studies suggest exacerbation of negative effects for daughters and for behaviorally inhibited children. However, unanswered questions about the context of parenting and appropriate comparison groups limit generalization from the current research base. The impact of timing or sequencing also has been neglected. Yet, it seems likely that sequencing—the ordering of onset of mental illness and childbirth or episodes of mental illness and child developmental phases—will critically color both parenting and outcomes for children. A number of authors have also proposed that episodes of mental illness during infancy are likely to undermine the mother’s capacity to build positive interactions with her child. This lack of positive maternal structure is said to increase risk of poor attachment and undermine children’s sense of competence (Hammen, 1991a). Because context clearly matters, it may be critical to compare how mothers with similar life circumstances cope with long-term and episodic chronic physical illnesses, in order to determine the extent to which findings are specific to mental illness. The likelihood that many mentally ill women will raise their children makes it important to draw on the current empirical literature to extrapolate appropriate assessment and intervention applications. Because women with a mental illness are likely to have children, their existence, age, and whereabouts should be one of the initial foci of assessment. When children are involved, plans should be developed for their stable care, and parenting should be bolstered by appropriate supports. In our own research (Oyserman, Bybee, Mowbray, & McFarlane, 2000), we have found that current stresses and functioning, rather than diagnosis and provision of social supports to mothers, is particularly important in predicting positive parenting attitudes and behavior (e.g., nurturance and involvement with children’s education). Careful assessment of the course of the mother’s mental illness, its interplay with childbirth and child rearing, and detailed attention to the current expression of the illness may provide a road map for intervention—although the current cross-sectional research base does not offer much to
Annual progress in child psychiatry and child development 2000–2001
186
build on. Because their children are at risk of exhibiting social, emotional, and behavioral problems at home and school, the parenting challenges these mother face should be assessed and appropriate supports provided, including interventions with mothers and children. Mothers are more likely to be effective when given the opprotunity to take on responsibilities (Goepfert, Webster, Pollard, & Nelki, 1996), yet they need ongoing support in parenting to reduce the negative sequelae of past and future episodes of mental illness. It seems reasonable to conclude that mothers’ capacity to parent will be framed by their past and current experiences and life situations, and not simply by their diagnostic category. Research to date has not provided a window into the way that women’s past life experiences—not only mental illness, but related stresses such as poverty, victimization, and involvement in mental health services—are incorporated into maternal identities, shaping their current focus, motivation, skills, and abilities. Within this general framework, it is clear that current and ongoing supports can bolster women’s efforts to parent, just as current and ongoing stressors can siphon their energies and attention (Klehr et al., 1983; Stott et al., 1984). Though few studies have focused directly on mothers’ maternal identity, some research suggests that mothers with a mental illness find parenting to be an important and fulfilling aspect of their lives (Oyserman, Bybee, Mowbray, & Khang, 2000; Perkins, 1992). Though mothers may exhibit problematic parenting practices, they may well be interested in and amenable to interventions that facilitate increased sense of maternal selfcompetence and efficacy. Future research must build on previous work by utilizing larger, more representative samples of women; following the women and their children over time; and analyzing the impact of demographic, material-economic, and social-relational support and stress factors, as well as diagnosis, on women’s sense of themselves as mothers. The direct and indirect effects of these factors on parenting and outcomes for children must be studied, as well. Only by means of much larger samples can such analyses proceed; without these more costly study designs, our ability to make sense of the ways in which serious mental illness influences parenting and outcomes for children will be severely impaired. Research that addresses more adequately the current gaps in our knowledge can be the basis for interventions that provide culturally appropriate support networks for these women—not as people with mental illness, but as mothers who must guide and support their growing children and carry out the tasks of parenting in particularly difficult circumstances.
ACKNOWLEDGMENTS This work was supported in part by NIMH grants 5R01MH5432105, 1R01MH5749501 Al to the first and second authors, and a W.T.Grant Faculty Scholar Award to the first author. Authors are at the Institute for Social Research, University of Michigan, Ann Arbor.
Parenting among mothers with a serious mental illness 187
REFERENCES Alexander, M., & Muenzenmaier, K. (1998). Trauma, addiction, and recovery. In B. L.Levin, A.K., Blanch, & A.Jennings (Eds.), Women’s mental health services: A public health perspective (pp. 215–239). Thousand Oaks, CA: Sage. Amminger, G.P., Pape, S., Rock, D., Roberts, S.A., Ott, S.L., Squires-Wheeler, E., Kestenbaum, C., & Erlenmeyer-Kimling, L. (1999). Relationship between childhood behavioral disturbance and later schizophrenia in the New York High-Risk Project. American Journal of Psychiatry, 156, 525–530. Andrews, B., Brown, G., & Creasey, L. (1990). Intergenerational links between psychiatric disorders in mothers and daughters: The role of parenting experiences. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 31, 1115–1129. Belle, D. (Ed) (1982). Lives in stress: Women and depression, Beverly Hills: Sage. Belle, D. (1990). Poverty and women’s mental health. American Psychologist, 45, 385– 389. Campbell, S., Cohn, J., & Meyers, T (1995). Depression in first-time mothers: Motherinfant interactions and depression chronicity. Developmental Psychology, 31, 349–357. Caton, C., Cournos, F., Felix, A., & Wyatt, R. (1998). Childhood experiences and current adjustment of offspring of indigent patients with schizophrenia. Psychiatric Services, 49, 86–90. Cohler, B., Gallant, D., Grunebaum, H., Weiss, J., & Gamer, E. (1980). Child-care attitudes and development of young children of mentally ill and well mothers. Psychological Reports, 46, 31–46. Cohler, B., & Musick, J. (1983). Psychopathology of parenthood; Implications for mental health of children. Infant Mental Health Journal, 4(3), 140–163. Cohler, B., Stott, F., & Musick, J. (1996). Distressed parents and their young children: Interventions for families at risk. In M.Goepfert (Ed.), Parental psychiatric disorder: Distressed parents and their families (pp. 107–134). Cambridge, UK: Cambridge University Press. Cohn, J.F., Matias, R., Tronick, E.Z., Connell, D., & Lyons-Ruth, K. (1986). Face to face interactions of depressed mothers and their infants. New Directions for Child Development , 34, 31–45. Cohn, J.F., & Tronick, E.Z. (1987). Mother-infant face-to-face interaction: The sequence of dyadic states at 3, 6, and 9 months. Developmental Psychology, 23, 68–77. Cohn, J.F., & Tronick, E.Z. (1988). Mother-infant face-to-face interaction: Influence is bidirectional and unrelated to periodic cycles in either partner’s behavior. Developmental Psychology, 24, 386–392. Cohn, J.F., & Tronick, E.Z. (1989). Specificity to infants’ response to mothers’ affective behavior. Journal of American Academy of Child and Adolescent Psychiatry, 57, 663– 667. Conrad, M., & Hammen, C. (1989). Role of maternal depression in perceptions of child maladjustment. Journal of Consulting and Clinical Psychology, 152, 663–667.
Annual progress in child psychiatry and child development 2000–2001
188
Cooper, P., Campbell, E., Day, A., Kennerley, H., & Bond, A. (1988). Nonpsychotic psychiatric disorder after childbirth. British Journal of Psychiatry, 152, 799–806. Cox, A., Puckering, C., Pound, A., & Mills, M. (1987). The impact of maternal depression in young children. Journal Child Psychiatry, 28, 917–928. Davenport, Y., Zahn-Waxler, C., Adland, M., & Mayfield, A. (1984). Early child-rearing practices in families with a manic-depressive parent. American Journal of Psychiatry, 141, 230–235. Davies, P., & Windle, M. (1997). Gender-specific pathways between maternal depressive symptoms, family discord, and adolescent adjustment. Developmental Psychology, 33, 657–668. DeChillo, N., Matorin, S., & Hallahan, C. (1987). Children of psychiatric patients: Rarely seen or heard. Health and Social Work, 12, 296–302. Dennis, D., Buckner, J., Lipton, F., & Levine, I. (1991). A decade of research and services for homeless mentally ill persons: Where do we stand? American Psychologist, 46, 1129–1138. Downey, G., & Coyne, J. (1990). Children of depressed parents: An integrative review. Psychological Bulletin, 108, 50–76. Erickson, E. (1968). Identity, youth and crisis. New York: Norton. Field, T. (1984). Early interactions between infants and their postpartum depressed mothers. Infant Behavior and Development, 7, 517–522. Field, T. (1986). Models for reactive and chronic depression in infancy. In E.Tronick & T.Field (Eds.), Maternal depression and infant disturbance: New directions for child development (pp. 47–60). San Francisco: Jossey-Bass. Field, T., Estroff, D., Yando, R., del Valle, C., Malphurs, J., & Hart, S. (1996). “Depressed” mothers’ perceptions of infant vulnerability are related to later development. Child Psychiatry and Human Development, 27, 43–53. Field, T., Healy, B., Goldstein, S., & Gutherz, M. (1990). Behavior-state matching and synchrony in mother-infant interactions of nondepressed versus depressed dyads. Developmental Psychology, 26, 7–14. Field, T., Healy, B., & Leblanc, W.G. (1989). Sharing and synchrony of behavior states and heart rate in nondepressed versus depressed mother-infant interactions. Infant Behavior and Development, 12, 357–376. Frankel, K.A., & Harmon, R.J. (1996). Depressed mothers: They don’t always look as bad as they feel. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 289–298. Free, K., Alechina, I., & Zahn-Waxler, C. (1996). Affective language between depressed mothers and their children: The potential impact of psychotherapy. Journal of the American Academy of Child & Adolescent Psychiatry, 35, 783–790. Gaensbauer, T., Harmon, R., Cytryn, L., & McKnew, D. (1984). Social and affective development in infants with a manic-depressive parent. American Journal of Psychiatry, 141, 223–229. Garb, H. (1997). Race bias, social class, bias, and gender bias in clinical judgment. Clinical Psychology: Science & Practice, 4, 99–120. Gelfand, D., & Teti, D. (1990). The effects of maternal depression on children. Clinical Psychology Review, 10, 329–353.
Parenting among mothers with a serious mental illness 189 Gelfand, D., Teti, D., & Fox, C. (1992). Sources of parenting stress for depressed and mondepressed mothers of infants. Journal of Clinical Child Psychology, 21, 262–272. Ghodsian, M., Zajicek, E., & Wolkind, S. (1984). A longitudinal study of maternal depression and child behavior problems. Journal of Child Psychology and Psychiatry, 25, 91–109. Gizynski, M. (1985). The effects of maternal depression on children. Clinical Social Work Journal, 13, 103–116. Goepfert, M., Webster, J., Pollard, J., & Nelki, S. (1996). The assessment and prediction of parenting capacity: A community-oriented approach. In M.Goepfert (Ed.), Parental psychiatric disorder: Distressed parents and their families (pp. 271–309). Cambridge, UK: Cambridge University Press. Gonzales, L., Kelly, J., Mowbray, C., Hays, R., & Snowden, L. (1991). Community mental health. In M.Hersen, A.Kazdin, & A.S.Bellack (Eds.), The clinical psychology handbook. New York: Pergamon Press. Goodman, S., Adamson, L., Riniti, J., & Cole, S. (1994). Mothers’ expressed attitudes: Associations with maternal depression and children’s self-esteem and psychopathology. Journal of the American Academy of Child & Adolescent Psychiatry, 33, 1265–1274. Goodman, S.H., & Brumley, H.E. (1990). Schizophrenic and depressed mothers: Relational deficits in parenting. Developmental Psychology, 26, 31–39. Goodman, S.H., & Johnson, M.S. (1986). Life problems, social supports, and psychological functioning of emotionally disturbed and well low-income women. Journal of Community Psychology, 14, 150–158. Gordon, D., Burge, D., Hammen, C., & Adrian, C. (1989). Observations of interactions of depressed women with their children. American Journal of Psychiatry, 146, 50–55. Greenberg, M.T. (1998). Fast Track: A comprehensive model for promoting competence in childhood. Paper presented at the annual meeting of the American Psychological Association, San Francisco. Gross, D. (1983). How some dyads “fail”: A qualitative analysis with implications for nursing practice. Infant Mental Health Journal, 4, 272–286. Hamilton, E.B., Jones, M., & Hammen, C. (1993). Maternal interaction style in affective disordered, physically ill, and normal women. Family Process, 32(3), 329–340. Hammen, C. (1991a). Depression runs in families: The social context of risk and resilience in children of depressed mothers. New York: Springer-Verlag. Hammen, C. (1991b). Generation of stress in the course of unipolar depression. Journal of Abnormal Psychology, 100, 555–561. Hammen, C. (1997). Children of depressed parents: The stress context. In S.Wolchik (Ed.), Handbook of children’s coping: Linking theory and intervention (pp. 1 31–157). New York: Plenum Press. Hammen, C., Burge, D. & Adrian, C. (1991). Timing of mother and child depression in a longitudinal study of children at risk. Journal of Consulting and Clinical Psychology, 59, 341–345. Hammen, C., Burge, D., & Stansbury, K. (1990). Relationship of mother and child variables to child outcomes in a highrisk sample: A causal modeling analysis. Developmental Psychology, 26, 24–30.
Annual progress in child psychiatry and child development 2000–2001
190
Hammen, C., Gordon, D., Burge, D., Adrian, C., Jaenicke, C., & Hiroto, D. (1987.) Maternal affective disorders, illness and stress: Risk for children’s psychopathology. American Journal of Psychiatry, 144, 736–741. Harnish, J., Dodge, K., & Valente, E. (1995). Mother-child interaction quality as a partial mediator of the roles of maternal depressive symptomology and socioeconomic status in the development of child behavior problems. Child Development, 66, 739–753. Harris, J. (1995). Where is the child’s environment? A group socialization theory of development. Psychological Review, 102, 458–489. Hirshfeld, D., Biederman, J., Brody, L., Faraone, S., & Rosenbaum, J. (1997). Expressed emotion toward children with behavioral inhibition: Associations with maternal anxiety disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 910–917. Hops, H., Biglan., A., Sherman, L., Arthur, J., Friedman, L., & Osteen, V. (1987). Home observations of family interactions of depressed women. Journal of Consulting & Clinical Psychology, 55, 341–346. Inoff-Germain, G., Nottelmann, E., & Radke-Yarrow, M. (1992). Evaluative communications between affectively ill and well mothers and their children. Journal of Abnormal Child Psychology, 20, 189–212. Inoff-Germain, G., Nottelmann, E., & Radke-Yarrow, M. (1997). Relation of parental affective illness to family, dyadic, and individual functioning: An observational study of family interaction. American Journal of Orthopsychiatry, 67, 433–448. Jacob, T., & Johnson, S. (1997) Parent-child interaction among depressed fathers and mothers: Impact on child functioning. Journal of Family Psychology, 11, 391–409. Jameson, P., Gelfand, D., Kulcsar, E., & Teti, D. (1997). Mother-toddler interaction patterns associated with maternal depression. Development and Psychopathology, 9, 537–550. Kessler, R.C. (1997). The effects of stressful life events on depression. Annual Review of Psychology, 48, 191–214. Klehr, K.B., Cohler, B.J., & Musick, J.S. (1983). Character and behavior in the mentally ill and well mother. Infant Mental Health Journal, 4, 250–270. Kochanska, G., Kuczynski, L., Radke-Yarrow, M., & Welsh, J. (1987). Resolutions of control episodes between well and affectively ill mothers and their young children. Journal of Abnormal Child Psychology, 15, 441–456. Lundy, B., Field, T., Cigales, M., & Cuadra, A. (1997). Vocal and facial expression matching in infants of mothers with depressive symptoms. Infant Mental Health Journal, 18, 265–273. Lundy, B., Field, T., McBride, C., Abrams, S., & Carraway, K. (1997). Child psychiatric patients’ interactions with their mothers. Child Psychiatry & Human Development, 27, 231–40. Lundy, B., Field, T., & Pickens, J. (1996). Newborns of mothers with depressive symptoms are less expressive. Infant Behavior and Development, 19, 419–424. Luster, T., & Okagaki, L. (Eds.) (1993). Parenting: An ecological perspective. Hillsdale, NJ: Erlbaum. Markus, H., & Cross, S. (1990). The interpersonal self. In L.Pervin (Ed.), Handbook of personality: Theory and research (pp. 576–608). New York: Guilford.
Parenting among mothers with a serious mental illness 191 McNeil, T., Kaij, L., Malmquist-Larsson, A., Naeslund, B., Persson-Blennow, I., McNeil, N., & Blennow, G. (1983). Offspring of women with nonorganic psychoses. Acta Psychiatrica Scandinavica, 68, 234–250. McNeil, T., Naeslund, B., Persson-Blennow, I., & Kaij, L. (1985). Offspring of women with nonorganic psychosis: Mother-infant interaction at three-and-a-half and six months of age. Acta Psychiatrica Scandinavica, 71, 551–558. Merikangas, K., & Spiker, D. (1982). Assortative mating among in-patients with primary affective disorder. Psychological Medicine, 12, 753–764. Mills, M., Puckering, C., Pound, A., & Cox, A. (1985). What is it about depressed mothers that influence their children’s functioning? In J.E.Stevenson (Ed.), Recent research in developmental psychopathology (pp. 11–17). Oxford, UK: Pergamon Press. Mowbray, C.T., Oyserman, D., Lutz, C., & Pumell, R. (1996). Women: The ignored majority. In L.Spaniol (Ed.), Psychological and Social Aspects of Psychiatric Disability (pp. 171–194). Boston: Center for Psychiatric Rehabilitation, Sargent College of Allied Health Professions, Boston University. Mowbray, C.T., Oyserman, D., & Ross, S. (1995). Parenting and the significance of children for women with a serious mental illness. Journal of Mental Health Administration, 22, 189–200. Mowbray, C.T., Oyserman, D., Zemencuk, J., & Ross, S. (1995). Motherhood for women with serious mental illness: Pregnancy, childbirth and the postpartum period. American Journal of Orthopsychiatry, 65, 21–38. Musick, J., Stott, F., Spencer, K., Goldman, J., & Cohler, B. (1984). The capacity for “enabling” in mentally ill mothers. Zero to Three, 4, 1–6. Naeslund, B., Persson-Blennow, I., McNeil T., & Kaij, L. (1985). Offspring of women with nonorganic psychosis: Mother-infant interaction at three and six weeks of age. Acta Psychiatrica Scandinavica, 71, 441–450. Naeslund, B., Persson-Blennow, I., McNeil T., Kaij, L., & Malmquist-Larsson, A. (1984). Offspring of women with nonorganic psychosis: Infant attachment to the mother at one year of age. Acta Psychiatrica Scandinavica, 69, 231–241. Nicholson, J., Nason, M., Calabresi, A., & Yando, R. (1999). Fathers with severe mental illness. American Journal of Orthopsychiatry, 69, 134–141. Nicholson, J., Sweeney, E.M., & Geller, J.L. (1998a). Mothers with mental illness: I. The competing demands of parenting and living with mental illness. Psychiatric Services, 49, 635–642. Nicholson, J., Sweeney, E.M., & Geller, J.L. (1998b). Mothers with mental illness: II. Family relationships and the context of parenting. Psychiatric Services, 49, 643–649. Nolen-Hoeksema, S., Wolfson, A., Mumme, D., & Guskin, K. (1995). Helplessness in children of depressed and nondepressed mothers. Developmental Psychology, 31, 377– 87. Oates, M. (1989). Management of major mental illness in pregnancy and the puerperium. Bailliere’s Clinical Obstetrics and Gynaecology, 3, 905–920. Olson, S., & Banyard, V. (1993). “Stop the world so I can get off for a while”: Sources of daily stress in the lives of lowincome single mothers of young children . Family Relations, 42, 50–56.
Annual progress in child psychiatry and child development 2000–2001
192
Oyserman, D., Benbenishty, R., & Ben Rabi, D. (1992). Characteristics of children and their families at entry into foster care. Psychiatry and Human Development. 22, 199– 211. Oyserman, D., Bybee, D., Mowbray, C.T., & Khang, S. (2000). The meaning of motherhood for women with mental illness. Unpublished manuscript, University of Michigan, Ann Arbor. Oyserman, D., Bybee, D., Mowbray, C.T., & McFarlane, P. (2000). Support helps: Effects of community functioning and social support on parenting of African American mothers with a serious mental illness. Unpublished manuscript, University of Michigan, Ann Arbor. Oyserman, D., Mowbray, C.T., & Zemencuk, J.K. (1994). Resources and supports for mothers with a severe mental illness. Health and Social Work, 19, 132–142. Patterson., G.L. (1998). Growth in new forms of antisocial behavior: On becoming a system. Paper presented at the annual meeting of the American Psychological Association, San Francisco, CA. Perkins, R. (1992). Catherine is having a baby. Feminism and Psychology, 2, 110–112. Persson-Blennow, I., Binett, B., & McNeil, T.F. (1988). Offspring of women with nonorganic psychosis: Antecedents of anxious attachment to the mother at one year of age Acta Psychiatrica Scandinavia 78, 66–71. Persson-Blennow, I., McNeil, T.F., & Blennow, G. (1988). Temperament in children delivered by vacuum extraction. American Journal of Orthopsychiatry, 58, 304–309. Persson-Blennow, I., Naeslund, B., McNeil T.F., & Kaij, L. (1986). Offspring of women with nonorganic psychosis: Mother-infant interaction at one year of age. Acta Psychiatrica Scandinavia, 73, 207–213. Pound, A., Cox, A., Puckering, C., & Mills, M. (1985). The impact of maternal depression on young children. In J.E.Stevenson (Ed.), Recent research in developmental psychopathology. Oxford, UK: Pergamon Press. Puckering, C. (1989). Annotation: Maternal depression. The Journal of Child Psychology & Allied Disciplines, 30, 807–17. Quinton, D., Rutter, M., & Liddle, C. (1984). Institutional rearing, parenting difficulties and marital support. Psychological Medicine, 14, 107–124. Radke-Yarrow, M., Cummings, E., Kuczynski, L., & Chapman, M. (1985). Patterns of attachment in two and three year olds in normal families and families with parental depression. Psychological Bulletin, 112, 485–499. Radke-Yarrow, M., McCann, K., DeMulder, E., Belmont, B., Martinez, P., & Richardson, D. (1995). Attachment in the context of high-risk conditions. Development and Psychopathology, 7, 247–265. Radke-Yarrow, M., Nottelmann, E., Belmont, B., & Welsh, J.D. (1993). Affective interactions of depressed and nondepressed mothers and their children. Journal of Abnormal Child Psychology, 21, 683–695. Radke-Yarrow, M., Zahn-Waxler, C., Richardson, D.T., Susman, A., & Martinez, P. (1994). Caring behavior in children of clinically depressed and well mothers. Child Development, 65, 1405–1414. Reid, J.B. (1998). Proximal and distal impacts of a conduct problem prevention intervention. Paper presented at the annual meeting of the American Psychological
Parenting among mothers with a serious mental illness 193 Association, San Francisco, CA. Richards, J.P. (1990). Postnatal depression: A review of recent literature. British Journal of General Practice, 40, 472–476. Rieder, R.O. (1973). The offspring of schizophrenic parents: A review. The Journal of Nervous and Mental Disease, 157, 179–190. Rogosch, F.A., Mowbray, C.T., & Bogat, G.A. (1992). Determinants of parenting attitudes in mothers with severe psychopathology. Development and Psychopathology, 4, 469–487. Romans-Clarkson, S., Walton, V., Herbison, G., & Mullen, P. (1988). Marriage, motherhood and psychiatric morbidity in New Zealand. Psychological Medicine, 18, 983–990. Ruttenberg, W., Finello, K., & Cordeiro, A. (1997). Interactions between depressed and nondepressed Latina mothers and their premature infants. Infant Mental Health Journal, 18, 364–377. Rutter, M., Silberg, J., O’Connor, T., & Simonoff, E. (1999). Genetics and child psychiatry: II Empirical research findings. Journal of Child Psychology and Psychiatry, 40, 19–55. Sameroff, A.J., Seifer, R., & Barocas, R. (1983). Impact of parental psychopathology: Diagnosis, severity, or social status effects? Infant Mental Health Journal, 4, 236–249. Sands, R.G. (1995). The parenting experience of low-income single women with serious mental disorders. Families in Society, 76, 86–96. Scherer, D., Melloh, T., Buyck, D., Anderson, C., & Foster, A. (1996). Relation between children’s perception of maternal mental illness and children’s psychological adjustment. Journal of Clinical Child Psychology, 25(2), 156–169. Schwab, B., Clark, R.E., & Drake, R.E. (1991). An ethnographic note on clients as parents. Psychosocial Rehabilitation Journal, 15(2), 95–99. Seifer, R., Sameroff, A., Anagnostopolou, R., & Elias, P. (1992). Mother-infant interaction during the first year: Effects of situation, maternal mental illness, and demographic factors. Infant Behavior and Development, 15, 405–426. Stein, A., Gath, D., Bucher, J., Bond, A., Day, A., & Cooper, P. (1991). The relationship between post-natal depression and mother-child interaction. British Journal of Psychiatry, 158, 46–52. Stott, F.M., Musick, J.S., Clark, R., & Cohler, B.J. (1983). Developmental patterns in the infants and young children of mentally ill mothers. Infant Mental Health Journal, 4, 217–234. Stott, F.M., Musick, J.S., Cohler, B.J., Spencer, K.K., Goldman, J., Clark, R., & Dincin, J. (1984). Intervention for the severely disturbed mother. In J.Musick & B.Cohler (Eds.), Intervention Among Psychiatrically Impaired Parents and Their Young Children (pp. 7–32). San Francisco: Jossey-Bass. Tableman, B. (1987). Stress Management Training: An approach to the prevention of depression in low-income populations. In R.Munoz (Ed.), Depression prevention: Research directions (pp. 171–184). Washington, DC: Hemisphere. Tannenbaum, L., & Forehand, R. (1994). Maternal depressive mood: The role of the father in preventing adolescent problem behaviors. Behavior Research & Therapy, 32, 321–325.
Annual progress in child psychiatry and child development 2000–2001
194
Tarullo, L., DeMulder, E., Martinez, P., & Radke-Yarrow, M. (1994). Dialogues with preadolescents and adolescents: Mother-child interaction patterns in affectively ill and well dyads. Journal of Abnormal Child Psychology, 22, 33–51. Tarullo, L., DeMulder, E., Ronsaville, D., Brown, E., & Radke-Yarrow, M. (1995). Maternal depression and maternal treatment of siblings as predictors of child psychopathology. Developmental Psychology, 31(3), 395–405. Teti, D., & Gelfand, D. (1991). Behavioral competence among mothers of infants in the first year: The mediational role of maternal self-efficacy. Child Development, 62, 918– 929. Teti, D., Gelfand, D., Isabella, R., & Messinger, D. (1995). Maternal depression & the quality of early attachment: An examination of infant preschoolers and their mothers. Developmental Psychology, 31, 364–76. Teti, D., Gelfand, D., & Pompa, J. (1990). Depressed mothers’ behavioral competence with their infants: Demographic and psychosocial correlates . Development and Psychopathology, 2, 259–270. Tronick, E. (1989). Emotions and emotional communication in infants. American Psychologist, 44, 112–119. Tronick, E.Z., & Gianino, A.F. (1986a). Interactive mismatch and repair: Challenges to the coping infant. Zero to Three, 6, 1–6. Tronick, E.Z., & Gianino, A.F. (1986b). The transmission of maternal disturbance to the infant. New Directions for Child Development, 34, 5–11. Waldo, M.C., Roath, M., Levine, W., & Freedman, R. (1987). A model program to teach parenting skills to schizophrenic mothers. Hospitals and Community Psychiatry, 38 (10), 1110–1112. Wang, A., & Goldschmidt, V. (1994). Interviews of psychiatric inpatients about their family situation and young children. Acta Psychiatrica Scandinavia, 90, 459–465. Waters, B.G.H., & Marchenko-Bouer, I. (1980). Psychiatric illness in the adult offspring of bipolar manic-depressives. Journal of Affective Disorders, 2, 119–126. Webster-Stratton, C., & Hammond, M. (1988). Maternal depression and its relationship to life stress, perceptions of child behavior problems, parenting behaviors and child conduct problems. Journal of Abnormal Child Psychology, 16, 299–315. Zahn-Waxler, C., Iannotti, R., Cummings, E., & Denham, S. (1990). Antecedents of problem behaviors in children of depressed mothers. Development and Psychopathology, 2, 271–291. Zlochower, A., & Cohn, J. (1996) Vocal timing in face-to-face interaction of clinically depressed and nondepressed mothers and their 4-month-old infants. Infant Behavior and Development, 19, 371–374.
Part III ATTENTION DEFICITHYPERACTIVITY DISORDER
PART III: ATTENTION DEFICIT-HYPERACTIVITY DISORDERS
10 Diagnostic Efficiency of Neuropsychological Test Scores for Discriminating Boys With and Without Attention Deficit-Hyperactivity Disorder Alysa E.Doyle, Joseph Biederman, Larry J.Seidman, Wendy Weber, and Stephen V.Faraone
A growing literature has documented group differences between boys with and without attention deficit-hyperactivity disorder (ADHD) on neuropsychological tests; however, whether or not such tests can discriminate individuals with ADHD from non-ADHD controls remains unclear. This study used conditional probability and receiver operating characteristic analyses to examine the efficiency of testbased diagnostic discriminations in a large sample of referred boys with and without ADHD. Single neuropsychological tests had limited discriminating ability at various cutoff scores. When multiple tests were used together, prediction of ADHD status improved but overall diagnostic efficiency remained limited. Diagnostic efficiency did not differ when medicated and nonmedicated index children were considered separately. Results suggest that children with ADHD show variable deficits on neuropsychological tests of attention and executive functions. Impairments on multiple neuropsychological tests are predictive of ADHD, but normal scores do not rule out the diagnosis. The prognostic implications of variable neuropsychological deficits in children with ADHD require further investigation.
Alysa E.Doyle and Joseph Biederman, Department of Psychiatry, Harvard Medical School at Massachusetts General Hospital; Larry J.Seidman and Stephen V.Faraone, Department of Psychiatry, Harvard Medical School at Massachusetts General Hospital, and Massachusetts Mental Health Center, Boston, Massachusetts; Wendy Weber, Pediatric Psychopharmacology Unit, Massachusetts General Hospital, Boston, Massachusetts.
Annual progress in child psychiatry and child development 2000–2001
198
INTRODUCTION Recent evidence supports the hypothesis that attention deficit-hyperactivity disorder (ADHD) is a developmental brain disorder with primary deficits in the frontal cortex and/or subcortical (e.g., striatal) regions projecting to the frontal lobes (Voeller, 1990). Symptoms of ADHD, including difficulties with motoric inhibition, sustained attention, and behavioral regulation, are similar to those seen in adults with lesions of the prefrontal cortex (Mattes, 1980). Furthermore, recent neuroimaging studies of children and parents of children with ADHD point to abnormalities in both structure and function of the frontal regions that regulate attentional and motor intentional behavior (Castellanos et al., 1994; Filipek, Semrud-Clikeman, Steingrad, Kennedy, & Biederman, 1997; Giedd et al., 1994; Semrud-Clikeman et al., 1994; Zametkin et al., 1990). Consistent with the hypothesis that frontal deficits represent the neural substrate of ADHD is a growing literature revealing deficits in children with ADHD as compared with controls on neuropsychological tests presumed to assess frontal lobe functions. This literature, which is primarily composed of studies examining preadolescent boys, indicates that children with ADHD manifest subaverage or relatively weak performance on various tasks of vigilance and sustained attention, motoric inhibition, executive functions (such as organization and complex problem solving), and verbal learning and memory (Barkley, 1997; Pennington & Ozonoff, 1996). Although the early studies on this topic show slight inconsistencies (Barkley, 1990), such discrepancies are likely due to methodological issues and do not detract from the overall finding of group differences between children with ADHD and normal and psychiatric controls (Barkley & Grodzinsky, 1994). More recent studies that have used relatively larger samples, more rigorous diagnostic procedures, and a wider variety of neuropsychological tasks (e.g., Barkley, Grodzinsky, & DuPaul, 1992; Grodzinsky & Diamond, 1992; Seidman, Biederman, Faraone, Weber, & Ouellette, 1997) also indicate group differences. Our work (Seidman et al., 1997), as well as that of others (e.g., Fischer, Barkley, Edelbrock, & Smallish, 1990), extends these findings from children to adolescents with ADHD. Our recent data further suggest that neuropsychological impairment in individuals with ADHD cannot simply be explained by comorbid psychiatric (i.e., anxiety, depressive, or conduct disorders) or learning disorders (Seidman, Biederman, et al., 1995). In addition, children with ADHD do not appear to be impaired on simple motor speed or visuospatial accuracy, thus suggesting executive impairment is not a symptom of a generalized deficit (Seidman et al., 1997). Given the strong association between the diagnosis of ADHD and attentionalexecutive deficits, clinicians have speculated that neuropsychological tests could be used to aid in the diagnostic classification of children with ADHD (Barkley & Grodzinsky, 1994; Culbertson & Krull, 1996; Goldstein & Goldstein, 1998; Grodzinsky & Diamond, 1992; Rutter, Tuma, & Lann, 1988). However, group differences alone are insufficient indices of a test’s predictive utility (Baldessarini, Finklestein, & Arana, 1983; Barkley & Grodzinsky, 1994; Elwood, 1993). Rather, analyses of conditional probabilities, including examinations of sensitivity, specificity, and positive and negative predictive
Diagnostic efficiency of neuropsychological
199
power (PPP and NPP, respectively), are more appropriate statistical tools for discerning the efficiency of test-based diagnostic discriminations. Despite the obvious scientific and clinical importance of this issue, the extent to which neuropsychological test scores can discriminate cases of ADHD has not yet been adequately examined in the literature. To date, only one study (Barkley & Grodzinsky, 1994) has used conditional probabilities to explore the efficiency of tests of attention and executive functioning in the classification of children with ADHD. In this investigation, although group differences between children with and without ADHD were found, neuropsychological tests had limited predictive utility for classifying individual cases as ADHD. Although certain tests showed acceptable PPP, all of the tests examined showed poor NPP. That is, although abnormal scores were indicative of ADHD, normal scores on the tests did not indicate the absence of this disorder. Although this study marks a valuable contribution to the neuropsychological literature on ADHD, its small sample size of approximately 12 participants per cell may limit the generalizability of these findings. Furthermore, this investigation assessed the conditional probabilities associated with single tests, whereas in clinical practice, the convergence of findings from multiple tests is preferable for assessing neuropsychological deficits (Seidman, 1997; Weiss & Seidman, 1988). Moreover, the extent to which a test scar should deviate from the norm in order for it to discriminate ADHD from non-ADHD cases is a question that was not addressed. The present study used an empirical approach to examine the utility of neuropsychological tests for discriminating children with ADHD from non-ADHD controls. On the basis of a careful review of the literature on the neuropsychology of ADHD, we constructed a battery of tests of attention and executive functioning to assess our large sample of well-characterized pediatrically and psychiatrically referred boys with and without Diagnostic and Statistical Manual of Mental Disorders-defined ADHD (3rd ed., rev.; DSM-III-R; American Psychiatric Association, 1987). We examined the diagnostic efficiency of single and multiple neuropsychological tests at three different cutpoints that represented various levels of impairment. Because the overall level of diagnostic efficiency was an empirical question to be answered by our analyses, we did not make specific predictions regarding how well the tests would discriminate cases from noncases. We did, however, anticipate that using multiple neuropsychological tests as a battery would provide better diagnostic discrimination than individual tests used alone and that more severe cutoffs would be superior to less severe ones.
METHOD Participants The present analyses were derived from the 4-year follow-up of a longitudinal study of ADHD (Biederman et al., 1992; Biederman et al., 1993; Faraone et al., 1992; Faraone, Biederman, Krifcher Lehman, et al., 1993; Faraone, Biederman, Sprich, et al., 1993). Detailed study methodology is reported elsewhere (Biederman et al., 1992; Biederman, Faraone, Milberger, Guite, et al., 1996). Briefly, the original sample consisted of 260 probands (140 ADHD and 120 control children) and their 822 first-degree biological
Annual progress in child psychiatry and child development 2000–2001
200
relatives. All of the participants were assessed at baseline with the structured interviews and intelligence and achievement measures described in the following. Probands and siblings did not receive direct assessments again until year 4. Neuropsychological tests of frontal lobe functioning were administered for the first time at the 4-year follow-up. Of the original 260 probands, 237 returned at year 4: analyses of this follow-up sample revealed no attrition biases (Biederman, Faraone, Milberger, Curtis, et al., 1996). Neuropsychological data were obtained on 95% of the returning probands (for ADHD children, N=123; for normal comparisons, N=103). This sample has been described previously (Seidman, Benedict, et al., 1995; Seidman, Biederman, et al., 1995; Seidman et al., 1997). Two independent sources—one psychiatric and one pediatric—provided the index children. We selected psychiatrically referred ADHD probands from consecutive referrals to the Pediatric Psychopharmacology Clinic at the Massachusetts General Hospital. This site is not a tertiary care clinic because approximately 50% of new referrals have never been diagnosed or treated before. Parents, pediatricians, and schools had referred these children for psychiatric evaluations. The pediatrically referred ADHD probands consisted of pediatric patients from the Harvard Community Health Plan, a large health maintenance organization. Within each setting, we also selected normal controls from active outpatients at pediatric medical clinics. At entry to the study, participants were Caucasian, non-Hispanic boys between 6 and 17 years of age. Potential participants were excluded if they had been adopted or if their nuclear family was not available for study. We also excluded participants with major sensorimotor handicaps (e.g., paralysis, deafness, and blindness), psychosis, autism, or a Full Scale IQ (FSIQ) estimate of less than 80. Participants from the lowest socioeconomic class (Hollingshead, 1975) were also excluded to reduce etiologic heterogeneity. Because such participants are exposed to poverty and extreme adversity, it is possible that their ADHD might have an etiology that differed substantially from other cases. All of the children characterized as ADHD met DSM-III-R diagnostic criteria for current ADHD at the time of the clinical referral (i.e., they had at least eight active symptoms of the disorder) according to their mothers’ report. Comparisons of children with and without ADHD (see Table 10.1) revealed small differences across demographic variables. ADHD children were slightly younger than controls. They were also lower than non-ADHD controls on grade level, socioeconomic status (SES), and IQ. Because this was a naturalistic study of referred ADHD children, 66% (N=81) of the 123 ADHD participants were receiving medications, whereas none of the controls were medicated. Of those medicated, 64% (N=52) were taking stimulants; 42% (N=34) were taking tricyclic antidepressants; and 14% (N=11) were taking other psychotropic medications, including Prozac, Zoloft, Haldol, and/or lithium. Measures Diagnostic data on children 12 years of age or older were obtained from independent interviews with them and their mothers using the Schedule for Affective Disorders and Schizophrenia for School-Age Children—Epidemiologic Version (K-SADS-E; Orvaschel & Puig-Antich, 1987). Children younger than 12 were not interviewed directly:
Diagnostic efficiency of neuropsychological
201
diagnostic information on these participants came from parent interviews only. All of the parents signed a written consent form for participation in the study. Interviewers had undergraduate degrees in psychology and were trained to high levels of interrater reliability. We computed kappa coefficients of agreement by having three experienced, board-certified child and adult psychiatrists diagnose participants from audiotaped interviews made by the interviewing staff. Based on 173 interviews, the median kappa for all of the K-SADS-E diagnoses was 0.86 and the kappa for ADHD was 0.99.
TABLE 10.1 Demographic Characteristics of the Sample
ADHD (N=113) Control (N=103) Demographic Variable M SD M SD Test Statistic p< Age (years) 14.7 3.0 15.3 3.8 t(224)=3.85 .001 Grade level 8.8 2.8 9.9 3.7 t(224)=2.62 .01 SES 1.8 1.0 1.4 0.7 z=−2.60 .01 Full Scale IQ 107.1 16.7 115.4 15.6 t(224)=2.62 .01 Note. ADHD=attention deficit-hyperactivity disorder; SES=socioeconomic status. We scored the ADHD diagnosis as positive if, based on the interview results, DSM-IIIR criteria were unequivocally met. All diagnostic uncertainties were resolved by a committee of four board-certified child and adult psychiatrists, chaired by Joseph Biederman, who were blind to the participants’ ascertainment group, ascertainment site, neuropsychological data, and all of the data collected from other family members. For children older than 12, the committee combined data from direct and indirect interviews by considering a diagnostic criterion positive if it was endorsed in either interview. Neuropsychological Tests We estimated FSIQ from the Vocabulary and Block Design subtests of the Wechsler Intelligence Scale for Children-Revised (WISC-R; Wechsler, 1974) for children up to and including age 16 and the Wechsler Adult Intelligence Scale-Revised (WAIS-R; Wechsler, 1981) for individuals 17 and older (Sattler, 1988). Academic achievement was assessed with the Arithmetic and Reading tests of the Wide Range Achievement TestRevised (WRAT-R; Jastak & Jastak, 1985). We used the procedure recommended by Reynolds (1984) and others (e.g., Frick et al., 1991), which is based on a statistically corrected discrepancy between IQ and achievement scores to define learning disabilities. The neuropsychological battery we developed was based on the empirical and clinical literatures on attention, executive functions, and ADHD. As stated, data reported in the present study were obtained from participants who returned for the fourth year of our longitudinal study, which marked the first time neuropsychological assessments were conducted on this sample. We assessed domains of functioning thought to be indirect indices of frontostriatal systems and thus important in ADHD. These include vigilance and distractibility, planning and organization, response inhibition, set shifting and categorization, selective attention, visual scanning, and verbal learning and memory.
Annual progress in child psychiatry and child development 2000–2001
202
Measures from several well-known clinical instruments were used. We used the ReyOsterrieth Complex Figure (ROCF; Osterrieth, 1944; Rey, 1941) to measure planning, organization, and visual memory; the Wide Range Assessment of Memory and Learning (WRAML; Adams & Sheslow, 1990) for children younger than 17 and the California Verbal Learning Test (CVLT; Delis, Kramer, Kaplan, & Ober, 1987) for individuals older than 17 to measure verbal list learning; the computerized Wisconsin Card Sorting Test (WCST; Grant, Ilai, Nussbaum, & Bigler, 1990; Heaton, Chelune, Talley, Kay, & Curtiss, 1993) to assess set-shifting and categorization; the Stroop Color-Word Test (Golden, 1978) to assess response inhibition; and the Scattered Letter Version of the Visual Cancellations Test A (Weintraub & Mesulam, 1985) to assess selective attention and visual scanning. We also chose an auditory continuous performance test (CPT; Weintraub & Mesulam, 1985) to measure vigilance, because this measure elicited cerebral metabolic abnormalities in ADHD adults (Zametkin et al., 1990) and because our preliminary data suggested that this version of the CPT was a useful task (Seidman, Biederman, et al., 1995). In addition, we calculated the Freedom From Distractibility Index (FFDJ) from the WISC-R and the WAIS-R using the Digit Span. Digit Symbol/Coding, and Arithmetic subtest scores (Wechsler, 1974). Although we included the Finger Tapping Test (Reitan & Wolfson, 1985) in our overall battery to measure motor speed, this task was excluded from the present analyses because it is not a direct measure of attention or executive functions. Thus, in the present analyses, subscales and indices from seven different neuropsychological measures were examined. Tests were administered in the following order: the WRAT-R Arithmetic and Reading; the ROCF Copy Task; the WISC-R/WIAS-R Vocabulary and Digit Span subtests; the ROCF Recall Task; the WISC-R/WAIS-R Block Design, Arithmetic, and Coding/Digit Symbol subtests; the Finger Tapping Test; the auditory CPT; the WRAML or the CVLT; the WCST; the Stroop ColorWord Test; and the Scattered Letter Version of the Visual Cancellations Task. The ROCF was administered and scored according to the methods described by Waber and Holmes (1985, 1986). We chose this developmental scoring system because it provides a specific measure of organizational strategy, whereas other scoring methods essentially assess visual spatial ability. Protocols were scored by two PhD-level clinical psychologists who were blind to all participant characteristics, including diagnosis. These individuals had been trained and supervised in the scoring method by one of this system’s originators (Jane Holmes Bernstein) and had achieved a high level of agreement with the supervisor (ranging from 94% to 100%) for all four test variables. For administration, the figure was reproduced such that the base rectangle measured 8.0 cm× 5.5 cm. All of the tests were administered and, with the exception of the ROCF, scored by neuropsychological technicians who had substantial training and supervision by Larry J.Seidman. To facilitate comparability across the WRAML and the CVLT list-learning measures, we calculated the percentage of words learned across all five trials of the tests for each participant and used this percentage as the verbal-learning score in the present analyses. Technicians were blind to the diagnosis of participants. Testing took only 2 to 2.5 hours. The battery was generally engaging and not overly taxing for participants, giving us a high degree of confidence that these results were not unduly affected by fatigue.
Diagnostic efficiency of neuropsychological
203
Plan of Analysis We estimated conditional probabilities using the measures of sensitivity, specificity, PPP, NPP, and total predictive value (TPV). As shown in Table 10.2, sensitivity reflects the probability of an abnormal test score given that a person has the diagnosis in question and specificity represents the probability of a normal test score given that the person does not have the diagnosis. To ascertain the probability of a diagnosis being present or absent given a certain test score, such as one might do in a clinical testing situation, measures of predictive power, which take into account the prevalence rate of a diagnosis in the population assessed, must also be considered (Baldessarini et al., 1983; Elwood, 1993; Meehl & Rosen, 1955). PPP indicates the probability of having the disorder given that a person has an abnormal test result, and NPP reflects the probability of not having the disorder given a normal test result. For a test to pick out individuals with a given diagnosis from a sample, both PPP and NPP must be adequate (i.e., greater than chance). Thus, a test’s TPV, which is based on both PPP and NPP, can be used as a general index of overall diagnostic efficiency. It is important to note, however, that tests can be clinically useful if only one of these indices is high. For example, if its PPP were high and its NPP were low, a test would be useful for confirming a diagnosis but not for ruling it out.
TABLE 10.2 Conditional Probability Analysis: Terminology
Term Sensitivity
Definition The probability of an abnormal test score given that a person has the diagnosis in question Specificity The probability of a normal test score given that the person does not have the diagnosis Positive predictive The probability of having the disorder given that a person has power an abnormal test result Negative predictive The probability of not having the disorder given a normal test power result Total predictive The probability of identifying individuals correctly for both value normal and abnormal test results
To calculate sensitivity, specificity, PPP, NPP, and TPV, it was necessary to designate cut scores that distinguished “normal” from “abnormal” performance. For each subscale or index; we used cutoffs of 1, 1.5, and 2 SDs above or below the mean of the 103 controls for that measure, depending on which direction represented imparied performance. For subscales that were not normally distributed, cutoffs were created at the percentile equivalents of 1, 1.5, and 2 SDs (i.e., ≥84.1%. ≥93.3%, and ≥97.7%, respectively, for those measures in which higher scores represented poor performance; and P15.9%, P6.7%, and P2.3% for those measures in which lower scores constituted poor performance). For purposes of discussion, we refer to cutoffs by their standard
Annual progress in child psychiatry and child development 2000–2001
204
deviation equivalents in the remainder of the article. We then examined sensitivity, specificity, PPP, NPP, and TPV for multiple tests in the battery considered together at each of the three severity cutoffs. To avoid overrepresenting measures with multiple subtests, we considered a test abnormal if it had at least one subcale that exceeded the threshold being examined. Given that conditional probability estimates are affected by base rates of the disorder in question, we also examined odds ratios (ORs) for single and multiple tests. ORs represent the odds of an abnormal test score in cases diagnosed with ADHD divided by the odds of the abnormal score in non-ADHD cases. Thus, the OR provides a general index of association; however, in contrast to the TPV, ORs are independent of the base rate of the diagnosis that is being examined. Finally, we used receiver operating characteristic (ROC) analyses (Murphy et al., 1987) to depict the trade-off between true positives and false positives that results from the range of scores on each individual subscale. At any threshold of a continuous scale (i.e., for our purposes, at each test score of a subscale or index score), the distance of the curve from the 45δ diagonal (which represents equal numbers of false positives and false negatives) represents the predictive strength of that threshold: the greater the distance from the curve, the better the score’s predictive ability. The area under the curve (AUC; Hanley & McNeil, 1982), which varies between 0.5 (no better than chance) and 1.0 (perfect prediction), is thus an overall index of predictive utility and adds to the present analyses by assessing diagnostic efficiency across all the cutoffs of a given scale rather than just the three severity cutoffs discussed in the preceding. We also calculated AUCs for the battery of tests as a whole, using the number of abnormal tests at the three different cutoffs as continuous scales.
RESULTS Group Differences on Neuropsychological Tests In a previous article (Seidman et al., 1997), we reported clear group differences in ADHD children versus non-ADHD controls on tests of attentionalexecutive functions in the present sample of boys. We review these findings briefly because of their relevance to the present analyses. ADHD boys were significantly more impaired than controls on six of the seven neuropsychological measures examined (and on 11 of the 15 possible individual subscales assessed). Given that multiple subscales were examined, we considered probabilities of p=0.01 as representing statistical significance in order to reduce the chance of committing a Type 1 error. Specifically, differences were found on the Word, Color, and Color-Word subscales of the Stroop ColorWord Test; the perseverative and nonperseverative error variables and the number of categories completed from the WCST; the Omissions subscale of the auditory CPT; the Disorganized Strategy rating on the Letter Cancellation Task: the Copy and Delay Organization scores from the ROCF; and the WISC-R/ WAIS-R FFDI. In addition, nonsignificant trends emerged on the Stroop Interference T score (p=0.02) and the WRAML/CVLT list-learning measure (p=0.03). The only measures on which the ADHD
Diagnostic efficiency of neuropsychological
205
group did not show worse performance were the Late Responses scale from the auditory CPT (p=0.06) and the failure to maintain set variable from the WCST (p=0.10). (See Seidman et al., 1997, for the details of these analyses as well as for these results of corrections by age, SES, family history, psychiatric comorbidity, and learning disabilities.) Group differences in the Seidman et al. study were not corrected for 1Q because of the possibility that controlling for FSIQ would control for aspects of the cognitive process related to the testing. Diagnostic Efficiency of Neuropsychological Test Scores Single tests. Table 10.3 illustrates the conditional probabilities for predicting ADHD status using single measures at various cut-points. For purposes of discussion, we use a TPV of equal to or greater than 60% as a general criterion for diagnostic efficiency (Biederman et al., 1993). By this criterion, the cutoff of 1.0 SD below the controls’ mean provided the most efficient diagnostic profile across different tests. At this cutoff, the PPPs for individual subtests ranged from modest to strong (range=56 to 76, M=67.9) and NPPs ranged from poor to modest (range=46 to 55, M=51.0). Measures that best discriminated cases from noncases were the Stroop Color-Word Test (Word and ColorWord T scores) and the FFDI, which, it is important to note, is composed of three subtests. As with the measures of predictive power, the best combination of sensitivity and specificity occurred at the 1.0-SD cutoff. Although sensitivities were highest at the 1.0-SD cutoff, they were generally low. In contrast, specificities were high across all three impairment thresholds.
TABLE 10.3 Conditional Probabilities: Prediction of ADHD Using Single Neuropsychological Tests at Varying Cutoff Points
Cutoff of 1 SD Cutoff of 1.5 SDs Cutoff of 2 SDs Below the Mean of Below the Mean of Below the Mean of the Controls the Controls the Controls Measure SE SP PPP NPP TPV SE SP PPP NPP TPV SE SP PPP NPP TPV Stroop Color-Word Test 41 83 74 54 60 23 93 80 50 55 14 96 81 48 51 Word T score 27 84 67 49 53 21 89 70 49 52 11 97 81 48 50 Color T score Color-Word T 43 84 76 55 62 21 92 76 49 54 19 97 88 50 54 score Interference T 32 81 66 50 54 11 92 62 46 48 6 97 70 46 47 score WCST Categories 34 84 71 52 57 26 89 74 50 55 19 95 82 49 53 completed Perseverative 38 77 66 51 56 20 88 68 48 51 17 91 70 48 51 errors Nonperseverative 39 80 70 52 58 28 88 74 51 56 17 92 72 48 51
Annual progress in child psychiatry and child development 2000–2001
206
errors Failure to 37 76 64 50 54 18 84 58 46 48 4 92 38 45 44 maintain set WRAML/CVLT 84 68 50 54 16 93 74 48 51 2 98 50 46 46 2% learning score Auditory CPT Omissions 34 77 64 49 54 15 91 68 47 50 8 95 67 46 48 Late responses 20 81 56 46 48 7 89 42 44 44 4 92 38 45 44 Letter Cancellation Task Organizational 43 71 64 51 56 11 90 58 46 47 11 90 58 46 47 Strategy ROCF 37 84 73 53 58 21 92 76 49 54 15 97 86 49 53 Copy Organization Delay 27 84 66 49 44 20 87 65 48 50 20 87 65 48 50 Organization WISC-R/WAIS-R 41 83 74 54 60 26 92 80 51 56 14 97 85 49 52 FFDI Note, Numbers in bold typeface represent TPV equal to or greater than 60. ADHD=attention deficit-hyperactivity disorder; CPT=continuous performance test; CVLT=California Verbal Learning Test; FFDI=Freedom From Distractibility Index; NPP=negative predictive power; PPP=positive predictive power; ROCF= Rey-Osterreith Complex Figure; SE=sensitivity; SP=specificity; TPV=total predictive value; WAIS-R=Wechsler Adult Intelligence ScaleRevised; WCST= Wisconsin Card Sort Test; WISC-R=Wechsler Intelligence Scale for Children-Revised; WRAML=Wide Range Assessment of Memory and Learning. Multiple tests. Table 10.4 illustrates the sensitivity, specificity, PPP, NPP, and TPV at the three cutoff thresholds when the seven neuropsychological tests from our battery were considered together. For tests with multiple subscales, we considered a test abnormal if at least one subscale surpassed the cutoff. Combining neuropsychological tests in this way yielded better diagnostic efficiency than did single tests. However, diagnostic efficiency was limited at all three cutoffs, with the 1.0-SD threshold once again showing better discrimination than the two more severe cutoffs. The most robust combination of PPP and NPP values (i.e., the highest TPVs) at this cutoff were observed when abnormal scores occurred on two, three, and four of the seven neuropsychological tests. Our data also show that the PPP increased and the NPP decreased along with the number of abnormal tests in the battery. Furthermore, the use of multiple tests together yielded considerably higher sensitivities than did the prior analyses of single scales used alone.
Diagnostic efficiency of neuropsychological
207
TABLE 10.4 Conditional Probabilities: Prediction of ADHD Using Multiple Neuropsychological Tests at Varying Cutoff Points
Cutoff of 1 SD Below Cutoff of 1.5 SDs Cutoff of 2 SDs Below the Mean of the Below the Mean of the the Mean of the Controls Controls Controls SE SP PPP NPP TPV SE SP PPP NPP TPV SE SP PPP NPP TPV
No. of Abnormal Tests 1 92 19 58 67 59 74 44 61 58 60 59 63 65 56 61 2 76 46 63 62 75 44 82 74 55 61 25 90 76 50 55 3 57 72 71 58 64 28 92 81 52 57 15 95 79 49 52 4 42 88 81 56 63 14 95 77 48 51 7 96 69 46 48 5 26 93 82 51 57 7 99 90 47 49 3 99 80 46 47 6 11 98 87 48 50 3 99 80 46 47 1 100 100 46 46 7 5 99 86 47 48 1 99 50 46 46 0 100 100 46 46 Note. Numbers in bold typeface represent TPV equal to or greater than 60. ADHD=attention deficit-hyperactivity disorder; NPP=negative predictive power; PPP= positive predictive power; SE=sensitivity; SP=specificity; TPV=total predictive value. The low NPP found in the above analyses suggests variability of performance on neuropsychological tests for both the ADHD and non-ADHD samples, with some boys showing impaired scores on many tests and others showing no or few difficulties. Figure 10.1 illustrates the frequency of impaired neuropsychological tests in each group. We then considered the possibility that being on medication accounted for the relatively better performance of some members of the ADHD group on these tests. We examined this issue by comparing the number of abnormal test scores in the medicated (n=71) and nonmedicated (n=51) ADHD participants at the three impairment cutoffs. (The medication status of one participant was unknown.) Table 10.5 shows these comparisons.
TABLE 10.5 Number of Abnormal Neuropsychological Tests: Medicated Versus Nonmedicated Boys with ADHD
Nonmedicated (n=51) Medicated (n=71) Cutoff M SD Range M SD Range 1.0 SD 2.7 1.9 0–7 3.4 1.9 0–7 1.5 SDs 1.4 1.5 0–7 2.0 1.7 0–6 2.0 SDs 0.8 1.3 0–6 1.3 1.4 0–5 Note. ADHD=attention deficit-hyperactivity disorder.
z p −1.91 0.06 −2.18 0.03 −2.30 0.02
Annual progress in child psychiatry and child development 2000–2001
208
Figure 10.1. Variability within attention deficit-hyperactivity disorder (ADHD) and non-ADHD groups of performance on neuropsychological tests of executive functions: Number of tests impaired at 1 SD (A), number of tests impaired at 1.5 SDs (B), and number of tests impaired at 2 SDs (C). Individuals in both the medicated and nonmedicated groups exhibited a range of performance on the battery, with some showing no abnormal scores and others showing abnormal scores on all or nearly all of the tests administered. The number of impaired tests was slightly higher in the medicated as compared with the nonmedicated participants at all three cutoffs, although this difference did not reach our p=.01 criterion for statistical significance. Thus, these results suggest that present
Diagnostic efficiency of neuropsychological
209
psychopharmacological treatment was not associated with better performance on the neuropsychological test battery and thus does not appear to account for the fact that some boys with ADHD showed minimal neuropsychological impairment on testing. Although the difference between the groups did not reach statistical significance, given that medicated participants had a slightly higher number of abnormal test scores than nonmedicated participants, we decided to further examine the relationship of medication status and variability of performance within the ADHD group. To do so, we separately discriminated the medicated and nonmedicated participants from the controls and compared the conditional probabilities of the tests in the two groups. We based these comparisons on sensitivities because PPPs and NPPs would be artificially constrained and enhanced, respectively, by the different numbers of ADHD participants in the groups (i.e., the probability of having ADHD would be lower in the nonmedicated versus medicated group because there are fewer cases of ADHD in the nonmedicated group; similarly, the probability of not having ADHD would be higher in the nonmedicated sample). Also, the specificities of the groups would not differ because this value is based on the performance of the controls. Results of these analyses revealed that, at all three impairment cutoffs, the only test that exhibited different sensitivities between the medicated and nonmedicated groups was the Delay Organization of the ROCF. At the 1.0-SD cutoff, the sensitivities of the medicated and nonmedicated groups were 35 and 14, respectively (z=2.69, p=.007). Specificities for both groups were 84. The PPP and NPP were adjusted for the base rates of ADHD using Bayes’s theorem (Kraemer, 1992). Adjusted PPPs of the medicated and nonmedicated groups were 93 and 35, respectively; adjusted NPPs were both 56. At the 1.5and 2.0-SD cutoffs, conditional probabilities were identical. The sensitivities of the medicated and nonmedicated groups were 28 and 7, respectively (z= 3.02, p=0.003). Specificities for both groups were 88. Adjusted PPPs of the medicated and nonmedicated groups were 79 and 20, respectively; adjusted NPPs were both 58. In addition to medication status, we also examined whether the remission of ADHD between its initial diagnosis and the administration of the neuropsychological battery at follow-up could have contributed to the variability within the ADHD group. When we assessed the 19 individuals whose ADHD did not persist to the 4-year follow-up of the study, these individuals exhibited a wide range of performance (i.e., zero to six tests). At the 1.0- and 1.5-SD impairment cutoffs, remitted individuals had a mean number of abnormal tests that were slightly, but not significantly, lower than the means of the nonremitted individuals at these cutoffs (2.2±1.9 versus 3.2±1.9 at the 1.0-SD cutoff, z= 2.22, p=0.03, and 1.1±1.4 vs. 1.8±1.6 at the 1.5-SD cutoff, z=2.15, p= 0.03). At the 2.0SD cutoff, however, the mean number of tests of the remitted group was significantly lower than that of the nonremitted group (0.6±1.1 versus 1.2±1.4, z=2.58, p<0.01). ORs: Single and Multiple Tests To further explore the utility of single versus multiple neuropsychological tests at different cutoff thresholds, we calculated ORs (see Table 10.6). As stated, the OR is a measure of association. Specifically, it reflects the odds of an abnormal test score in cases diagnosed with ADHD divided by the odds of the abnormal score in non-ADHD cases.
Annual progress in child psychiatry and child development 2000–2001
210
For purposes of discussion, we considered ORs greater than 3.0 as having good discriminative power (Biederman et al., 1993). Results were consistent with the notion that the odds of abnormal test scores were higher in ADHD participants as compared with non-ADHD participants. For individual tests, the hightest ORs occurred for the Stroop Color-Word Test, the WCST, the ROCF, and the FFDI. Combinations of tests yielded high ORs at all three deviation thresholds.
TABLE 10.6 Odds Ratios: Prediction of ADHD Based on Individual and Multiple Subtest Scores
Cutoff 1.5 SDs
Measure 1.0 SD 2.0 SDs Stroop Color-Word Test 3.34** 4.04* 3.97 Word T score 1.99 2.24 3.93 Color T score 3.83*** 3.18* 7.66*** Color-Word T score Interference T score 1.92 1.40 2.10 WCST No. of categories 2.62* 2.94* 4.50** Perseverative errors 2.03 1.93 2.15 Nonperseverative errors 2.49* 3.01* 2.44 Failure to maintain set 1.80 1.18 0.50 2.09 2.66 0.83 WRAML/CVLT % learning score Auditory CPT Omissions 1.70 1.90 1.73 Late Responses 1.55 1.90 1.73 Letter Cancellation Task Organizational Strategy 1.70 1.90 1.73 ROCF 3.02** 3.18* 6.08** Copy Organization 1.85 1.67 1.67 Delay Organization WISC-R/WAIS-R FFDI Multiple tests 3.23*** 4.17** 5.34* 1 test below cutoff 0.51 0.70 1.33 2 tests below cutoff 0.44 1.62 2.11 ≥3 tests below cutoff 3.37*** 4.54*** 3.58 Note. ADHD=attention deficit-hyperactivity disorder; CPT=continuous performance test; CVLT =California Verbal Learning Test; FFDI=Freedom From Distractibility Index; ROCF=ReyOsterrieth Complex Figure; WAIS-R=Wechsler Adult Intelligence Scale-Revised; WCST= Wisconsin Card Sort Test; WISCR=Wechsler Intelligence Scale for Children-Revised; WRAML =Wide Range Assessment of Memory and Learning. *p=.01, **p=.001, *** p=.0001.
Diagnostic efficiency of neuropsychological
211
ROC Analyses: AUC Finally, we examined ROC curves in order to assess the trade-off between true and false positives along the range of scores within individual test scales. As stated previously, AUCs range from 0.5 (no better than chance) to 1.0 (perfect discrimination) and provide a measure of a scale’s predictive utility across all possible thresholds. Thus, the AUC is useful for comparing the overall discriminative accuracy of different tests without having to designate a single cutoff threshold (Murphy et al., 1987). In the literature, there is no gold standard for the level of AUC because the AUC presumes that the costs of false negatives and false positives are comparable, and yet optimal combinations of sensitivity and specificity vary depending on the relative costs and benefits of the decisions being made based on a given test (Chen, Faraone, Biederman, & Tsuang, 1994). In the present study, our goal was to minimize both false positives and false negatives in order to maximize diagnostic precision. Therefore, the higher the AUC, the better the scale’s diagnostic efficiency. AUCs for individual scales, shown in Table 10.7, ranged from 0.56 to 0.72 and were consistent with ORs and conditional probabilities in that the best scales were the Stroop Color-Word Test, the WCST, the ROCF, and the FFDI. The most predictive measures were the FFDI, with an AUC of 0.72 (see Figure 10.2), and the entire battery considered as a whole (i.e., the number of impaired tests from zero to seven) using the 1.0-SD cutoff, with an AUC of 0.69. Because we found group differences between boys with and without ADHD on FSIQ (see Table 10.1), we repeated the preceding ROC analyses, correcting for IQ, in order to examine the possibility that that the group difference in intellectual ability was accounting for the discriminating power, although minimal, of the tests. These results, shown in Table 10.7. show slightly higher AUCs than those found when IQ was not corrected for, suggesting that the diagnostic efficiency of the tests was not merely the result of differential intellectual ability of the ADHD and non-ADHD groups.
TABLE 10.7 Area Under the Curve (AUC): Prediction of ADHD Using Individual and Combined Subtests
Measure Stroop Color-Word Test Word T score Color T score Color-Word T score Interference T score WCST No. of categories Perseverative errors Nonperseverative errors Failure to maintain set
AUC
AUC corrected for IQ
.6601 .6272 .6685 .5858
.7159 .6952 .7066 .6674
.6012 .6656 .6523 .5604
.6887 .6960 .7002 .6882
Annual progress in child psychiatry and child development 2000–2001
212
WRAML/CVLT % learning score .6011 .7037 Auditory CPT Omissions .6096 .6961 Late Responses .5711 .6834 Letter Cancellation Task .5911 .6895 Disorganized Strategy ROCF Copy Organization .6692 .7085 Delay Organization .6046 .6792 WISC-R/WAIS-R FFDI .7217 .7249 Combined tests 1.0-SD cutoff .6899 .7044 1.5-SD cutoff .6516 .6926 2.0-SD cutoff .6268 .6878 Note. ADHD=attention deficit-hyperactivity disorder; CPT=continuous performance test; CVLT =California Verbal Learning Test; FFDI=Freedom From Distractibility Index; ROCF=ReyOsterrieth Complex Figure; WAIS-R=Wechsler Adult Intelligence Scale-Revised; WCST= Wisconsin Card Sort Test; WISCR=Wechsler Intelligence Scale for Children-Revised; WRAML =Wide Range Assessment of Memory and Learning.
DISCUSSION The present study assessed the ability of neuropsychological tests of attention and executive functions to discriminate between boys with and without ADHD. Results showed that single neuropsychological tests had low sensitivity and limited discriminating ability. Using multiple tests as a battery resulted in only slightly improved diagnositic efficiency, with analyses of conditional probabilities revealing generally high PPP but only modest NPP. Thus, impairments on tests presumed to be indirect indices of frontostriatal systems supported the diagnosis of ADHD, but test results that were within normal limits could not rule out the diagnosis. These findings underscore the limited ability of these tests, even when used as a battery, to discriminate individuals with ADHD from normal controls without substantial false negatives. Optimal combinations of sensitivity, specificity, PPP, and NPP vary depending on the relative costs and benefits of decisions being made (Faraone & Tsuang, 1994). Although there are situations in which a screening tool that produces the present level of false negatives may be appropriate, the use of a lengthy procedure such as neuropsychological testing for discriminating cases from noncases presumes the minimization of both false positives and false negatives. Because conditional probability estimates are affected by base rates of the disorder in question, the modest diagnostic efficiency found in the present study may even be greater than would be found in samples with lower rates of ADHD. Furthermore, our results likely show better discrimination than would be found in clinical practice given that we aimed to distinguish boys with ADHD from normal,
Diagnostic efficiency of neuropsychological
213
rather than psychiatric, controls. Our finding of limited diagnostic efficiency can be attributed to the fact that, rather than showing uniform impairments, boys with ADHD exhibited a range of performance on tests of attention and executive functions. A matter for further research is the extent to which ADHD children with and without impaired performance on such tests differ in characteristics and outcome. Our data indicating the limited utility of single neuropsychological tests in diagnosing ADHD are consistent with the only other study to date that has used conditional probabilities to explore the role of tests of attentional-executive functions in the classification of children with and without this diagnosis (Barkley & Grodzinsky, 1994). Our study enhances the findings of Barkley and Grodzinsky by using multiple tests as a battery. That the use of multiple tests improves diagnostic efficiency is consistent with clinical practice. Neuropsychologists typically base conclusions on the convergence of evidence from several measures of a particular brain function (Seidman, 1997). Furthermore, because current conceptualizations of the networks that converge on and interact with the prefrontal cortex emphasize a multiplicity of functions, including attending, organizing, planning, sequencing, filtering, inhibiting responses and shifting mental set (Fuster, 1989), multiple tests may be better able to capture the complexity of these attentional-executive skills. However, even when the battery of tests was considered as a whole, our data revealed high PPP and only modest NPP. This limited NPP of neuropsychological tests for predicting ADHD reflects a range of performance within the ADHD and non-ADHD samples, with some boys performing in the impaired range on many tests and others showing no or few difficulties. It is this within-group variability that underlies the clinical implications of the present results. Neuropsychological tests are typically considered to have three broad uses (Seidman, 1997; Seidman & Toomey, 1999); (a) as aids to diagnosis; (b) as aids to treatment by identifying adaptive capacities, strengths, and limitations; and (c) as measures of change over time. The present study was designed to address the first application, and our results suggest that the present battery has limited utility for diagnosing ADHD. Nevertheless, our results also support the use of neuropsychological assessment for assessing strengths and weaknesses—the second goal described. Considering the fundamental importance of higher level cognitive functions, such as organization, planning, and synthesis of complex material, individuals with ADHD and more severe neuropsychological deficits may be at substantially higher risk for impaired educational, occupational, and interpersonal outcomes than individuals with ADHD who do not exhibit such severe deficits on testing. This is an empirical question that has yet to be answered. The reason behind the variability in ADHD children on neuropsychological tests is unclear. One possibility is that such variability is due to measurement error; however, the fact that we found variability across all of the tests makes this explanation unlikely. Another possible explanation is true diagnostic heterogeneity. That is, some children with ADHD may be more neuropsychologically compromised than others. Those who do not exhibit impaired test scores may have more subtle deficits that standard tests of attentionalexecutive functions may not be sensitive enough to elicit. Because individuals with executive deficits are highly responsive to external structure (Goldberg & Podell, 1995), it is possible that the structure imposed by the testing situation may, to some extent, mask less severe impairments. Such a phenomenon has been documented by
Annual progress in child psychiatry and child development 2000–2001
214
Draeger, Prior, and Sanson (1986), who found that performance on a CPT task deteriorates significantly in ADHD children but not in normal children when the examiner left the room (i.e., when the external regulator of behavior was removed). In the present study, the fact that some ADHD children exhibited poor performance on tests suggests more significant deficits in executive functions that emerged despite the highly structured testing situation.
Figure 10.2. The receiver operating characteristic (ROC) curve for the Wechsler Intelligence Scale for ChildrenRevised (WISC-R) Freedom From Distractibility Index (FFDI). The ROC curve shows the trade-off between true positives and false positives along the range of scores on a given scale. The area under the curve (AUC) is considered an index of the scale’s overall predictive utility. Depicted here is the ROC curve for the WISC-R FFDI, the measure in our battery with the highest AUC (.72). Because a large number of our participants were medicated, we also considered the possibility that some attentional deficits were ameliorated by medications. There is substantial evidence that stimulant medications improve performance on some neuropsychological measures, particularly measures of vigilance (Barkley, 1990), and, as stated, the majority of boys in the ADHD group were medicated at the time of assessment. Nevertheless, our analyses revealed generally comparable performance of the
Diagnostic efficiency of neuropsychological
215
medicated and nonmedicated ADHD participants on neuropsychological measures, with a trend toward medicated participants being more severely impaired. We also found that, with the exception of the Delay Organization scale of the ROCF, the sensitivities of the tests did not differ significantly when the medicated and nonmedicated groups were considered separately. Given that many ADHD children receive psychopharmacological treatment, the fact that treated samples continue to show neuropsychological deficits is relevant to many clinicians and educators. Further study of persons with ADHD before and after the use of stimulant and other medications is necessary to determine the impact of medications on test sensitivity. We emphasize, however, that our naturalistic study was not designed to assess the efficacy of medication in the treatment of neuropsychological deficits. Thus, inferences in this regard should be cautious given the well-known correlation between treatment and severity of illness (Faraone, Simpson, & Brown, 1992). Another important factor that could have contributed to the variability within the ADHD group is the remission of ADHD between its initial diagnosis and the administration of the neuropsychological battery at the 4-year follow-up. In a previous study (Biederman, Faraone, Milberger, Curtis, et al., 1996), we found that 85% of those with ADHD at baseline showed persistence of this disorder. For the present study, we determined that the 19 individuals whose ADHD did not persist ranged from having zero to six tests that were abnormal, with means that did not differ from those of the nonremitted individuals at the 1.0- and 1.5-SD impairment cutoffs. At the 2.0-SD cutoff, however, the mean number of tests of the remitted group was significantly lower than that of the nonremitted group. Thus, the remitted individuals did not account for the overall modest diagnostic efficiency found in our analyses, although remitted individuals were less likely to evince impairments at the most severe cutoff as compared with those whose ADHD did not remit. We also considered the possibility that the presence or absence of a learning disability could have accounted for the variable performance within our ADHD and non-ADHD groups because, at the time of testing, 23% of the ADHD participants and 6% of the nonADHD controls had either a math or reading disability. However, when we repeated all of the preceding analyses using only the non-learning disabled ADHD participants, our results did not change. (Results of these reanalyses are available from Alysa E.Doyle on request.) We must also consider the fact that our choice of tests may not have been optimal. The length and breadth of our assessment was limited by practical considerations. To test a large number of participants, we were compelled to use a relatively brief battery. Nonetheless, within such constraints, we chose tests that had a solid track record in the extant literature at the time we began this study in 1992. Given the same amount of time to test participants, the most likely alternative tests to those in our battery would be to substitute a more sensitive CPT than the one that we used. CPTs such as those described by Conners (1992) and others (e.g., Halperin, Sharma, Greenblatt, & Schwartz, 1991) have since had higher reported sensitivities than our CPT demonstrated. Another possibility is that tests of working memory would be particularly discriminating, inasmuch as some authors (e.g., Barkley, 1997) have suggested that working memory deficits are core impairments in ADHD.
Annual progress in child psychiatry and child development 2000–2001
216
Nonetheless, our battery of tests did have some discriminating power: Specifically, abnormal scores on the tests were suggestive of the ADHD diagnosis. Because we found group differences between the ADHD and non-ADHD boys on FSIQ, we examined whether this difference in intellectual ability accounted for the discrimination that was found. When we controlled for IQ in our ROC analyses, AUCs did not decrease but rather increased slightly, suggesting that confounded group differences on FSIQ did not account for the high PPP of the tests. Controlling for intellectual ability, however, is not a straightforward issue because, by doing so, one may be adjusting for the same aspects of the cognitive process that underlie the results of the neuropsychological testing. Thus, we did not attempt to control for FSIQ in our analyses of conditional probabilities. Our assessments of PPP and NPP were intended to serve as an empirical evaluation of the clinical practice of using neuropsychological tests to diagnose ADHD, and, in practice, clinicians are likely to base decisions on test results that are not controlled for IQ. Our overall findings must also be interpreted in light of certain statistical limitations. For example, our use of many test scores could have resulted in inflation of the Type I error rate. This limitation must be weighed against the careful selection of tests, which was based on a theoretical focus on the cognitive sequelae of frontal network dysfunctions, as well as on empirical evidence that these tests have shown group differences between ADHD and control children. As mentioned, another limitation relates to the fact that conditional probability estimates are affected by the base rates of the disorder in a sample. Therefore, the utility of this battery may differ in samples that have a different rate of ADHD. For this reason, we also examined ORs, which are independent of base rates, in our analyses. Nonetheless, it is important to note that the utility of these tests may be even more limited in samples with significantly lower rates of ADHD. To explore this issue further, our results should be cross-validated on other samples. A related point is that we used cutoff scores based on the means and standard deviations of our control group as opposed to population norms. Our decision to do so was based on an effort to compare boys with and without ADHD who were ascertained from the same population, thereby maximizing the statistical soundness of our procedures. It is noteworthy, though, that the controls in the present analyses generally had slightly higher scores on several of the measures as compared with population norms (e.g., the 1.0-SD cutoff for the Stroop Interference T score was 45 rather than 40). However, when we used the more stringent cutoffs of 1.5 SDs and 2.0 SDs below the means of controls, which may better approximate population norms, we also found limited diagnostic efficiency of our neuropsychological measures. A final limitation is that the order of administration of the neuropsychological tests was fixed rather than randomized. Thus, order effects may have gone undetected because of our design. However, that significant impairments in ADHD were as strong at the beginning of the evaluation (the ROCF) as at the end (the WCST and the Stroop ColorWord Test) Provides some evidence that the test results were not confounded by differential order effects or fa-tigue. Still, because there was no randomization, we cannot entirely rule out order effects as potential influences. Although the extent to which these results generalize to the DSM (4th ed.; DSM-IV; American Psychiatric Association, 1994) is unclear, the DSM-IV; field trials (Lahey et
Diagnostic efficiency of neuropsychological
217
al., 1994), as well as our own work (Faraone, Biederman, Weber, & Russell, 1998), indicate substantial overlap between DSM-III-Rand DSM-IV-defined ADHD. Furthermore, our research strongly indicates that the inattentive and combined subtypes in the DSM-IV represent quantitative differences in severity rather than qualitatively different entities (Faraone et al., 1998). Still, more work is needed to evaluate the neuropsychology of DSMIV subtypes. Despite these limitations, the present results have important implications. The DSMIII-R diagnosis of ADHD is based on an abnormal behavioral pattern of inattention, impulsivity, and hyperactivity, and neuropsychological tests of attention and executive functioning underidentify cases that meet these criteria. Nevertheless, neuropsychological testing may prove to be a helpful supplement to careful clinical interviewing, history, and informant observations by mapping out strengths and weaknesses that, for the individual child, would have important implications for treatment and educational remediation. Of course, the increased use of neuropsychological testing in the clinical assessment of persons with ADHD requires validation studies of the ecological utility of such measures. Considering the prohibitive expense in money and time of a full, comprehensive neuropsychological battery, there is a growing need to assess the utility of brief, “microbatteries” to answer focused questions (Milberg, 1996; Seidman, 1997). Thus, despite its lack of utility for diagnosing ADHD, this focal battery of tests of attentional and executive functions may have practical use in clinical settings.
ACKNOWLEDGMENTS This work was supported in part by U.S. Public Health Service Grant ROIMH-41314 from the National Institute of Mental Health. We thank J.Benjamin, R.Kolodny, and I.Kraus from the Pediatric Department of the Harvard Community Health Plan and James Perrin from the Pediatric Service of the Massachusetts General Hospital for their contribution to this work. We also thank Ken Benedict, Jane H.Bernstein, and Kari Seivard for their assistance.
REFERENCES Adams. W., & Sheslow, D. (1990). The Wide Range Assessment of Memory and Learning. Wilmington. DE: Jastak Assessments. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Baldessarini, R.J., Finklestein, S., & Arana, G.W. (1983). The predictive power of diagnostic tests and the effect of prevalence of illness. Archives of General Psychiatry, 40, 569–573. Barkley, R. (1997). Behavioral inhibition, sustained attention, and executive functions: Constructing a unifying theory of ADHD. Psychological Bulletin, 121, 65–94.
Annual progress in child psychiatry and child development 2000–2001
218
Barkley, R.A. (1990). Attention deficit hyperactivity disorder: A handbook for diagnosis and treatment. New York: Guilford. Barkley, R.A., & Grodzinsky, G.M. (1994). Are tests of frontal lobe functions useful in the diagnosis of attention deficit disorders? The Clinical Neuropsychologist, 8, 121– 139. Barkley, R.A., Grodzinsky, G., & DuPaul, G.J. (1992). Frontal lobe functions in attention deficit disorder with and without hyperactivity: A review and research report. Journal of Abnormal Child Psychology, 20, 163–188. Biederman, J., Faraone, S.V., Doyle. A., Lehman, B.K., Kraus, I., Perrin. J., & Tsuang, M.T. (1993). Convergence of the Child Behavior Checklist with structured interviewbased psychiatric diagnoses of ADHD children with and without comorbidity. Journal of Child Psychology and Psychiatry and Allied Disciplines, 34, 1241–1251. Biederman, J., Faraone, S.V., Keenan, K., Benjamin, J., Krifcher, B., Moore, C., Sprich, S., Ugaglia, K., Jellinek, M.S., Steingard, R., Spencer, T., Norman, D., Kolodny, R., Kraus, I., Perrin, J., Keller, M.B., & Tsuang, M.T. (1992). Further evidence for familygenetic risk factors in attention deficit hyperactivity disorder (ADHD): Patterns of comorbidity in probands and relatives in psychiatrically and pediatrically referred samples. Archives of General Psychiatry, 49, 728–738. Biederman, J., Faraone, S.V., Milberger, S., Curtis, S., Chen, L., Marrs, A., Ouellette, C., Moore, P., & Spencer, T. (1996). Predictors of persistence and remission of ADHD: Results from a four year prospective follow-up study of ADHD children. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 343–351. Biederman, J., Faraone, S., Milberger, S., Guite, J., Mick, E., Chen, L., Mennin, D., Marrs. A., Ouellette, C., Moore, P., Spencer, T., Norman, D., Wilens, T., Kraus, I., & Perrin, J. (1996). A prospective four-year follow-up study of attention deficit hyperactivity and related disorders. Archives of General Psychiatry, 53, 437–446. Castellanos, F., Giedd, J., Eckburg, P., Marsh, W., Vaituzis, C., Kaysen, D., Hamburger, S., & Rapoport, J. (1994). Quantitative morphology of the caudate nucleus in attention deficit hyperactivity disorder. American Journal of Psychiatry, 151, 1791–1796. Chen, W.J., Faraone, S.V., Biederman, J., & Tsuang, M.T. (1994). Diagnostic accuracy of the Child Behavior Checklist scales for attention deficit-hyperactivity disorder: A receiver operating characteristic analysis. Journal of Consulting and Clinical Psychology, 62, 1017–1025. Conners, C. (1992). Couriers’ Continuous Performance Test. Toronto, Ontario, Canada: Multi-Health Systems. Culbertson, J.L., & Krull, K.R. (1996). Attention deficit-hyperactivity disorder. In R.L.Adams, O.A.Parsons, J.L.Culbertson, & S.J.Nixon (Eds.), Neuropsychology for clinical practice: Etiology, assessment, and treatment of common neurological disorders (pp. 271–330). Washington, DC: American Psychological Association. Delis, D.C., Kramer, J.H., Kaplan, E., & Ober, B.A. (1987). California Verbal Learning Test-Adult Version. New York: Psychological Corporation. Draeger, S., Prior, M., & Sanson, A. (1986). Visual and auditory attention performance in hyperactive children: Competence or compliance. Journal of Abnormal Child Psychology, 14, 411–424. Elwood, R.W. (1993). Clinical discriminations and neuropsychological tests: An appeal
Diagnostic efficiency of neuropsychological
219
to Bayes theorem. Clinical Neuropsychologist, 7, 224–233. Faraone, S., Biederman, J., Chen, W.J., Krifcher, B., Keenan, K., Moore, C., Sprich. S., & Tsuang, M. (1992). Segregation analysis of attention deficit hyperactivity disorder: Evidence for single gene transmission. Psychiatric Genetics, 2, 257–275. Faraone, S., Biederman, J., Lehman, B.K., Keenan, K., Norman, D., Seidman, L., Kolodny, R., Kraus, I., Perrin, J., & Chen, W. (1993). Evidence for the independent familial transmission of attention deficit hyperactivity disorder and learning disabilities: Results from a family genetic study. American Journal of Psychiatry, 150, 891–895. Faraone, S.V., Biederman, J., Sprich, S., Buckminster, S., Chen, W., & Tsuang, M. (1993). Efficiency of diagnostic criteria for attention deficit disorder: Toward an empirical approach to designing and validating diagnostic algorithms. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 166–174. Faraone, S.V., Biederman, J., Weber, W., & Russell, R.L. (1998). Psychiatric, neuropsychological and psychosocial features of DSM-IV subtypes of attention deficit hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 185–193. Faraone, S.V., Simpson, J.C., & Brown, W.A. (1992). Mathematical models of complex dose-response relationships: Implications for experimental design in psychopharmacologic research. Staistics in Medicine, 11, 685–702. Faraone, S.V., & Tsuang, M.T. (1994). Measuring diagnostic accuracy in the absence of a gold standard. American Journal of Psychiatry, 151, 650–657. Filipek, P.A., Semrud-Clikeman, M., Steingrad, R., Kennedy, D., & Biederman, J. (1997). Volumetric MRI analysis: Comparing subjects having attention-deficit hyperactivity disorder with normal controls. Neurology, 48, 589–601. Fischer, M., Barkley, R.A., Edelbrock, C.S., & Smallish, L. (1990). The adolescent outcome of hyperactive children diagnosed by research criteria: II. Academic, attentional, and neuropsychological status. Journal of Consulting and Clinical Psychology, 58, 580–588. Frick, P.J., Lahey, B.B., Kamphaus, R.W., Loeber, R., Christ, M.A.G., Hart, E.L., & Tannenbaum, L.E. (1991). Academic underachievement and the disruptive behavior disorders. Journal of Consulting and Clinical Psychology, 59, 289–294. Fuster, J. (1989). The prefrontal cortex (2nd ed.), New York: Raven Press. Giedd, J.N., Castellanos, F.X., Casey, B.J., Kozuch, P., King, A.C., Hamburger, S. D., & Rapoport, J.L. (1994). Quantitative morphology of the corpus callosum in attention deficit hyperactivity disorder. American Journal of Psychiatry, 151, 665–669. Goldberg, E., & Podell, K. (1995). Later alligator in the frontal lobes. In H.H.Jasper, S.Riggio, & P.S.Goldman-Rakic (Eds.), Epilepsy and the functional anatomy of the frontal lobes (pp. 85–96). New York: Raven Press. Golden, C.J. (1978). Stroop Color and Word Test: A manual for clinical and experimental use, Chicago: Stoelting. Goldstein, S., & Goldstein, M. (1998). Managing attention deficit hyperactivity disorder in children: A guide for practitioners (2nd ed.). New York: Wiley. Grant, M.L., Ilai, D., Nussbaum, N.L., & Bigler, E.D. (1990). The relationship between continuous performance tasks and neuropsychological tests in children with attention-
Annual progress in child psychiatry and child development 2000–2001
220
deficit hyperactivity disorder. Perceptual and Motor Skills, 70, 435–445. Grodzinsky. G., & Diamond. R. (1992). Frontal lobe functioning in boys with attention deficit hyperactivity disorder. Developmental Neuropsychology, 8, 427–445. Halperin, J.M., Sharma, V., Greenblatt, E., & Schwartz, S. (1991). Assessment of the continuous performance test: Reliability and validity in a nonreferred sample. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 3, 603– 608. Hanley, J.A., & McNeil, B.J. (1982). The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 143, 29–36. Heaton, R.K., Chelune, G.J., Talley, J.L., Kay, G.G., & Curtiss, G. (1993). Wisconsin Card Sort Test manual (Rev.), Odessa, FL: Psychological Assessment Resources. Hollingshead, A.B. (1975). Four Factor Index of Social Status. New Haven, CT: Yale University, Department of Sociology. Jastak, J.F., & Jastak, S. (1985). The Wide Range Achievement Test-Revised. Wilmington, DE: Jastak Associates. Kraemer, H.C. (1992). Evaluating medical tests: Objective and quantitative guidelines. Newbury Park, CA: Sage. Lahey, B., Applegate, B., McBurnett, K., Biederman, J., Greenhill, L., Hynd, G., Barkley, R., Newcorn, J., Jensen, P., Richters, J., Garfinkel. B., Kerdyk, L., Frick, P., Ollendick, T., Perez, D., Hart, E., Waldman, I., & Shaffer, D. (1994). DSM-IV field trials for attention deficit hyperactivity disorder in children and adolescents. American Journal of Psychiatry, 151, 1673–1685. Mattes, J. (1980). A pilot trial of amantadine in hyperactive children. Psychopharmacology Bulletin, 16, 67–69. Meehl, P., & Rosen. A. (1955). Antecedent probability and the efficiency of psychometric signs, patterns, or cutting scores. Psychological Bulletin, 52, 194–216. Milberg, W.P. (1996). Issues in the assessment of cognitive functions in dementia. Brain and Cognition, 31, 114–132. Murphy, J., Berwick, D., Weinstein, M., Borus, J., Budman, S., & Klerman, G. (1987). Performance of screening and diagnostic tests: Application of receiver operating characteristic analysis. Archives of General Psychiatry, 44, 550–555. Orvaschel, H., & Puig-Antich, J. (1987). Kiddie SADS-Patient Version. Fort Lauderdale, FL: Nova University. Osterrieth, P.A. (1944). Le test de copie d’une figure complexe. Archives de Psychologie, 30, 206–256. Pennington, B.F., & Ozonoff, S. (1996). Executive functions and developmental psychopathology. Journal of Child Psychology and Psychiatry, 37, 51–87. Reitan, R.R., & Wolfson, D. (1985). The Halstead-Reitan Neuropsychology Test Battery: Theory and clinical interpretation. Tucson, AZ: Neuropsychology Press. Rey, A. (1941). L examen psychologique dans les cas d’encephalopathie traumatique. Archives de Psychologie, 28, 286–340. Reynolds. C.R. (1984). Critical measurement issues in learning disabilities. Journal of Special Education, 18, 451–476. Rutter. M., Tuma, A.H., & Lann, I.S. (Eds.) (1988). Assessment and diagnosis in child psychopathology. New York: Guilford.
Diagnostic efficiency of neuropsychological
221
Sattler, J.M. (1988). Assessment of children (3rd ed.). San Diego, CA: Author. Seidman L.J. (1997). Neuropsychological testing. In A.Tasman, J.Kay, & J.Lieberman (Eds.), Psychiatry (pp. 498–508). Philadelphia: W.B.Saunders. Seidman, L.J., Benedict, K., Biederman, J., Bernstein, J., Seiverd, K., Milberger, S., Norman, D., Mick, E., & Faraone, S. (1995). Performance of ADHD children on the Rey-Osterrieth Complex Figure: A pilot neuropsychological study. Journal of Child Psychology and Psychiatry, 36, 1459–1473. Seidman, L.J., Biederman, J., Faraone, S., Milberger, S., Norman, D., Seiverd, K., Benedict, K., Guite, J., Mick, E., & Kiely, K. (1995). Effects of family history and comorbidity on the neuropsychological performance of ADHD children: Preliminary findings. Journal of the American Academy of Child and Adolescent Psychiatry, 34, 1015–1024. Seidman, L.J., Biederman, J., Faraone, S.V., Weber, W., & Ouellette, C. (1997). Toward defining a neuropsychology of ADHD: Performance of children and adolescents from a large clinically referred sample. Journal of Consulting and Clinical Psychology, 65, 150–160. Seidman, L.J., & Toomey, R. (1999). The clinical use of psychological and neuropsychological tests. In. A.Nicholi (Ed.), The new Harvard guide to psychiatry (3rd ed., pp. 40–64). Cambridge, MA: Harvard University Press. Semrud-Clikeman, M., Filipek, P., Biederman, J., Steingard, R., Kennedy, D., & Renshaw, P. (1994). Attention deficit disorder: Differences in the corpus callosum and shape analysis in MRI morphometric analysis. Journal of the American Academy of Child and Adolescent Psychiatry, 33, 875–881. Voeller, K. (1990). The neurological basis of attention deficit hyperactivity disorder. International Pediatrics, 5, 171–176. Waber, D., & Holmes, J.M. (1985). Assessing children’s copy productions of the ReyOsterrieth Complex Figure. Journal of Clinical and Experimental Neuropsychology, 7, 264–280. Waber, D., & Holmes, J.M. (1986). Assessing children’s memory production of the ReyOsterrieth Complex Figure. Journal of Clinical and Experimental Neuropsychology, 8, 563–580. Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale for Children-Revised. New York: Psychological Corporation. Wechsler, D. (1981). Manual for the Wechsler Adult Intelligence Scale-Revised. San Antonio, TX: Psychological Corporation. Weintraub, S., & Mesulam, M.M. (1985). Mental state assessment of young and eld-erly adults in behavioral neurology. In M.M.Mesulam (Ed.), Principles of behavioral neurology (pp. 71–123). Philadelphia: F.A.Davis. Weiss, J.L., & Seidman, L.J. (1988). The clinical use of psychological and neuropsychological tests. In A.Nicholi (Ed.), The new Harvard guide to psychiatry (pp. 46–69). Cambridge, MA: Harvard University Press. Zametkin, A.J., Nordahl, T.E., Gross, M., King, C. Semple, W.E., Rumsey, J., Hamburger, S., & Cohen, R.M. (1990). Cerebral glucose metabolism in adults with hyperactivity of childhood onset. New England Journal of Medicine, 323, 1361–1366.
PART III: ATTENTION DEFICIT-HYPERACTIVITY DISORDERS
11 Efficacy of Methylphenidate among Preschool Children with Developmental Disabilities and ADHD Benjamin L.Handen, Heidi M.Feldman, Andrea Lurier, and Patty Jo Huszar Murray
Objective: This was a double-blind, placebo-controlled, crossover design study of the safety and efficacy of methylphenidate (MPH) in 11 preschool children (aged 4.0 to 5.11 years) with developmental disabilities and attention deficit-hyper activity disorder (ADHD). Method: MPH doses of .3 and .6 mg/kg per dose and a placebo were given. Drug response was evaluated via teacher-completed behavior checklists and clinic-based observations of activity level, attention, and compliance to adult requests. A side-effects checklist was also completed by teachers and parents. Results: Significant improvement on teacher ratings of hyperactivity and inattention as well as clinicbased observations of activity level and compliance were associated with MPH. Eight of 11 preschool children were medication responders (based on a minimum 40% decrease between placebo and one drug condition on either the teacher-rated Conners Hyperactivity Index or the HyperactiveDistractible subscale of the Preschool Behavior Questionnaire). Five children exhibited significant adverse drug side effects such as severe social withdrawal, increased crying, and irritability, especially at the higher dose (.6 mg/kg). Conclusions: Results suggest that preschool children with developmental disabilities and ADHD respond to MPH at rates similar to those of school-age children with mental retardation and ADHD. However, this population appears to be especially susceptible to adverse drug side effects (J Am Acad Child Adolesc Psychiatry, 1999 38(7):805–812. Key Words: attention deficit-hyperactivity disorder, methylphenidate, developmental disabilities, preschool children.
INTRODUCTION Attention deficit-hyperactivity disorder (ADHD) is characterized by significant deficits in
Efficacy of methylphenidate among preschool children
223
attention, impulse control, and activity level and affects 3% to 5% of the school-age population (Barkley, 1990). During the past 30 years, an extensive body of literature has documented the efficacy of stimulant medication in 75% to 80% of children with ADHD (Barkley, 1990). However, most of this research has focused on otherwise typically developing children 6 years of age and older. Only a small number of studies have examined stimulant medication effects among preschool children with ADHD. Estimates suggest that up to 5% of typically developing preschool children receive stimulants for management of problem behavior (Gadow, 1981). However, there has been some controversy regarding the safety and efficacy of stimulants in this population. For example, in a mid-1980s review of the literature, Rosenberg (1987) identified six published papers in the prior 15 years in which stimulant medication (methylphenidate) was used with typically developing preschool children with ADHD. With the exception of one study, he found nonsignificant results and deleterious side effects to be common. Even among children for whom positive outcomes were documented, results tended to be more variable and unpredictable than those reported among school-age children. On follow-up, many families chose to discontinue medication despite overall improvement in behavior during the medication trial. The prevailing view was that clinicians should take a cautious approach to the use of stimulants with preschool children given the lack of sufficient research in this area (Davy & Rodgers, 1989). A subsequent review of the literature in the early 1990s (Wilens & Biederman, 1992) cited data on a total of 130 preschool children with ADHD in four major studies (Barkley, 1988; Barkley et al., 1984; Conners, 1975; Schleifer et al., 1975). Taken together, these four studies reported few adverse drug side effects along with significant clinical improvement with the use of stimulants in preschool children. The authors concluded that these four studies disputed the prior notion that stimulants were ineffective and associated with adverse side effects in this population. The present consensus was recently summarized by Barkley (1998), who indicated that there are overwhelming data to support the efficacy of stimulants in children 5 years of age and older. However, the stimulant response rate is probably much less for children between 4 and 5 years of age, and stimulant use is not recommended for those under the age of 3 years. Historically, there has been a dearth of research among children with ADHD and developmental disabilities or mental retardation. Most studies have specifically excluded this group from serving as subjects in medication protocols. However, recent studies focusing on stimulant medication efficacy among school-age children with mental retardation and ADHD have documented positive MPH response rates ranging from 37% to 75% (Aman et al., 1991, 1993; Handen et al., 1990, 1992a, 1995). The overall response rate is thus less than the 75% to 80% positive clinical response rate documented in studies of typically developing children with ADHD (Barkley et al., 1993). In addition, Handen et al. (1991) found a greater number of adverse drug side effects among children with mental retardation and ADHD than had been reported among schoolage children without cognitive deficits. Studies of stimulant use in preschool children with developmental disabilities are limited to case reports despite estimates that stimulant use among preschool children with developmental disabilities is around 5.5% (Gadow, 1977). We know of only four papers
Annual progress in child psychiatry and child development 2000–2001
224
specifically detailing the effects of stimulants in this population. Shafto and Sulzbacher (1977) found a 5-mg dose of MPH to increase attention to tasks but to have variable effects on toy changes during a free play observation with a 4-year-old with mental retardation, language deficits, and emotional disturbance. Schell et al. (1986) documented increased compliance and decreased oppositional behavior with the use of MPH in a 5year-old with ADHD and an IQ of 60. Gadow and Pomeroy (1990) treated a 4year-old boy who had ADHD and an IQ of 45. Results found initial success with MPH, increased moodiness and whining when the medication regimen was switched to D-amphetamine, and a worsening of ADHD symptoms when MPH was reintroduced. Finally, Helsel et al. (1989) presented four single case studies of MPH in children with mental retardation; one subject, a 4.9-year-old with mild mental retardation, showed no improvement on behavior checklists, but some improvement in on-task behavior and work accuracy. A second subject, a 5-year-old with an IQ of 80 (low-average range of intelligence), evidenced a significant decrease on behavior checklist ratings of hyperactivity along with gains in ontask behavior and work accuracy. The issue of medicating preschool children with developmental disabilities and/or mental retardation is controversial. For children who are functioning 1 to 2 years below age expectancy, it is difficult to determine whether their activity levels and impulsive styles are consistent with their mental ages. The Physicians’ Desk Reference (1996) does not recommend the prescribing of methylphenidate (MPH) for children younger than 6 years of age. Yet for a small group of children the behaviors associated with ADHD may result in suspension from preschool or day care, severe family disruption, increased risk for physical abuse, and the potential need for inpatient psychiatric treatment. The purpose of this study was to extend our knowledge of the safety and efficacy of MPH among preschool children with developmental disabilities as a first step toward establishing guidelines for use. A primary objective was to examine the rate of positive responding to MPH among this group of children, who we hypothesized would respond at rates similar to the nondelayed preschool population. A second objective was to determine optimal doses for gains in attention and activity level. A third goal was to examine the impact of MPH on play and mother-child interactions. Finally, we sought to monitor and describe the rate and type of adverse medication side effects.
METHOD Subjects Eleven children, aged 4.0 to 5.11 years (mean 58.9 months, SD 8.2 months), served as subjects. Nine were male; IQs ranged from 40 to 78 (mean 60.0, SD 11.6). All subjects scored at or above the 90th percentile on both a teachercompleted Preschool Behavior Questionnaire (Behar & Stringfield, 1974) and the Hyperactivity Index of the Conners Parent Rating Scale (Goyette et al., 1978), which served as the inclusionary criteria for the study. In addition, all subjects had been previously evaluated by an interdisciplinary team of developmental specialists, during which time either a diagnosis of ADHD was confirmed or long-term concerns with inattention and overactivity were documented.
Efficacy of methylphenidate among preschool children
225
Nine of the 11 subjects met criteria for ADHD, based on a parent-completed questionnaire using DSM—III—R criteria (American Psychiatric Association, 1987). Two of these nine subjects also met criteria for oppositional defiant disorder, based on this same questionnaire. A diagnosis of autism/pervasive development disorder was ruled out for all subjects at the time of the interdisciplinary team assessment. IQ testing indicated receptive/expressive language functioning to be consistent with the overall IQ for the majority of subjects. There did not appear to be any relationship between IQ and severity of ADHD symptoms. All subjects were enrolled in developmental preschool programs at the time of the study. None had previously been prescribed stimulant medication. Setting Subjects were assessed in a 10×18-foot room equipped with a one-way mirror for observation. All sessions were videotaped. PROCEDURE The study involved a double-blind, crossover design with two doses of MPH (0.3 and 0.6 mg/kg per dose) and placebo. The first week of the study was devoted to taking baseline (off-drug) measures at each child’s preschool program. In addition, each child participated in an off-drug laboratory session to acclimate the subject to the new setting and obtain baseline data. During weeks 2 through 4 of the study, a dose of medication was given 30 to 60 minutes before the beginning of each child’s preschool program. For those children enrolled in a full-day program (n=7), a second dose was given with lunch. Two children received a third MPH dose in the midafternoon. Families contin-ued to give medication during the weekend. Drug-placebo order was randomly assigned. Each dose level was given for 7 consecutive days. However, the lower, 0.3 mg/kg dose always preceded the higher, 0.6 mg/kg dose (because of concerns regarding potential significant adverse side effects at the higher dose). Preschool staff, parents, and laboratory staff were unaware that the lower MPH dose preceded the higher MPH dose, thereby eliminating any potential bias. Preschool staff were contacted weekly to respond to questions and ensure compliance with the protocol. Weekly laboratory sessions also provided opportunities to meet with families to confirm protocol compliance.
DEPENDENT MEASURES Weekly measures of behavior and performance were obtained from the child’s family, preschool teacher, and laboratory sessions. However, family measures are not reported, as most families observed their children on medication for only relatively short periods of time (4 to 8 hours each Saturday and Sunday) and there was considerable variation in the exposure of family observers (e.g., a child might be with a mother, father, or grandparent on different days).
Annual progress in child psychiatry and child development 2000–2001
226
Preschool Classroom Measures The following materials were completed on the last day of each phase (weekly) so that teachers had the opportunity to observe subjects for the same period of time across phases. Conners Teacher Rating Scale. This is a 28-item behavior problem checklist with rating based on a four-point scale (0=“not at all” to 3=“very much”; Goyette et al., 1978). Four indices are obtained: 1) conduct problems; 2) hyperactivity; 3) inattention-passivity; and 4) hyperactivity index. The scale is normed for children aged 3 to 17 years. Preschool Behavior Questionnaire. This is a 30-item behavior problem checklist developed and normed for a preschool population (Behar & Stringfield, 1974). Items are rated on a 3-point scale (0=“not a problem” to 2=“very much”). Three indices are obtained: 1) hostile-aggressive; 2) anxious; and 3) hyperactive-distractible. Side Effects Checklist. This 18-item checklist (Handen et al., 1991) was completed by preschool teachers and parents. Items were derived from the list of possible side effects contained in the Physicians’ Desk Reference (1988). Severity of each symptom is rated on a seven-point Likert scale. Laboratory Measures The following measures were taken during weekly laboratory sessions conducted under each of the 3 drug conditions. A 10-second interval recording system was used to rate behaviors. Waiting Task. This involved placing the child at a table with a book and instructing him/her to sit and look at a book while the mother and evaluator talked (Kolmen et al., 1995). The adults discussed the directions for the session while seated a few feet behind the child. The mother was instructed to do whatever she would normally do when talking with another adult while her child was in the room (with the exception of holding the child in her lap). Coded behaviors include on-task behavior (looking at the book for at least 1 second), child initiated contact with parent, disruptive behavior (e.g., yelling, throwing objects, running out of room), and out-of-seat behavior. Resistance to Temptation. This task was designed to provide a measure of impulsivity. The child was seated at a table and a large, enticing toy was placed before him/her. The examiner began to show the child how the toy worked, then suddenly stopped, explaining that an important piece was missing. The child was told not to touch the toy while the examiner left the room to find the missing piece. The child was subsequently observed alone for 2 minutes. A different toy was used for each drug condition. Coded behaviors included toy touches, time to first touch, and out-of-seat behavior. Play Session. During an 8-minute play session, the mother and child were together in the room with 12 toys arranged on the floor. The mother was given a book to read and instructed to look at the book and to respond to the child only if approached. The child was told to play with the toys. Coded behaviors included play intensity, movement, disruptive behavior, and toy changes (i.e., when the child initiated or ceased to play with a toy). In addition, the span of time spent playing with the same toy(s) was assessed:
Efficacy of methylphenidate among preschool children
227
epoch 1: the number of intervals a child played with the same toy for 20 seconds or less; epoch 2: the number of intervals a child played with the same toy for 21 to 120 seconds; and epoch 3: the number of intervals a child played with the same toy for 120 seconds or more. Compliance Task. This was based on a task described by Handen et al. (1992b) and immediately followed the 8-minute play session. The mother and child remained in the room with toys. At 1-minute intervals the mother was signaled to give a simple one-step command to the child (e.g., “put the car on the table,” “give me a hug”). A total of 10 commands were given; three parallel sets of commands were randomly assigned across drug conditions. Coded be-haviors included compliance within 10 seconds of the request, disruptive behavior, and parental prompts, praise, and physical guidance. Clean-Up Task. This task was also designed to assess compliance with maternal requests. Immediately after the compliance task, the mother was instructed to ask her child to clean up the toys by placing them in a basket. The following behaviors were coded during the 6-minute observation: cleaning up (percentage of intervals child was observed placing a toy in the basket), disruptive behavior, vocalizations, and maternal praise and physical guidance.
RELIABILITY Inter-rater reliability was calculated on 16% of observation intervals for the waiting task, play session, compliance, and clean-up tasks. Subjects and sessions were randomly selected from the baseline and three drug phases. Two primary coders and a reliability coder were used. Percentage of agreement across coders ranged from 68.4% to 100% (mean of 92.6%) over the 19 primary coding domains used.
DATA ANALYSIS Data were analyzed using a multivariate repeated-measures analysis of variance (ANOVA) among the three drug conditions. Data for 10 of the 11 subjects were used in the analyses; the 11th subject’s data were incomplete because adverse side effects at the 0.3 mg/kg dose precluded the completion of the 0.6 mg/kg phase. Percentile data were transformed using an arcsine-square root transformation; numeric data were transformed using a square root transformation. All variables were on a continuous scale of measurement. Sphericity test results indicated that conditions for the univariate repeatedmeasures ANOVA were not met (Collier et al., 1972; Davidson, 1972). Where appropriate, a Scheffe post hoc procedure (nondirectional, 0.05 level) was used (Marascuilo & Levin, 1983).
RESULTS Results are presented for 10 of the 11 subjects who completed the protocol. The 11th
Annual progress in child psychiatry and child development 2000–2001
228
child experienced significant adverse side effects (see the following) and was not given the 0.6 mg/kg dose. Therefore, his data are not included in the data analysis. Data from teacher-completed behavior checklists across drug conditions are provided in Table 11.1. Scores on measures of inattention and activity level (e.g., Hyperactivity, InattentionPassivity, and Hyperactivity Index subscales of the Conners Teacher Rating Scale and the Hyperactive-Distractible subscale of Preschool Behavior Questionnaire) decreased significantly, indicating improvement between placebo and the 0.6 mg/kg MPH dose. Measures of conduct problems (based on the Conduct Problems subscale of the Conners and the Hostile-Aggressive subscale of the Preschool Behavior Questionnaire), although decreasing in the expected direction, failed to show a significant change. Anxiety was rated as evidencing no change across drug conditions. Mean baseline scores for both conduct problems measures and anxiety were low.
TABLE 11.1 Comparison of Methylphenidate and Placebo on Dependent Measures (N=10)
Placebo 0.3 mg/kg 0.6 mg/kg Measure Mean SD Mean SD Mean SD p Conners Teacher Rating Scale Conduct Problems 6.1 5.1 4.6 5.3 4.1 4.30.566 Hyperactivity 14.0 3.7 9.0 5.1 6.2 3.40.001** Inattention-Passivity 12.8 4.3 9.9 3.9 8.9 3.70.030* 17.4 6.0 11.9 5.7 9.2 3.80.006* Hyperactivity Index Preschool Behavior Questionnaire 7.3 4.9 4.7 4.7 4.5 5.10.283 Hostile-Aggressive Anxious 4.8 2.5 5.5 3.4 4.6 1.30.486 7.0 1.5 4.7 2.5 3.8 1.50.001** Hyperactive-Distractible Teacher-rated side effects No. reported 4.6 2.1 6.1 4.5 5.5 2.70.466 Severity rating 14.1 7.0 15.8 14.6 15.0 8.70.907 Waiting task On-task 1 second 23.8 22.4 31.2 27.6 29.8 35.00.715 Child-mother 28.1 25.2 31.2 26.2 20.9 36.00.116 Disruptive 29.4 29.8 21.8 18.9 24.3 33.40.643 Out of seat 80.8 22.5 73.3 32.8 53.1 45.90.057 Resistance to temptation First touch 3.2 5.3 3.2 5.3 5.0 6.00.164 % Touch 57.5 44.0 67.5 45.5 37.5 42.30.042a Out of seat 50.0 41.3 24.0 32.5 45.0 48.00.463 Play session 34.1 25.5 26.0 26.5 10.3 18.80.000** Intensity Movement 47.6 21.5 36.5 19.8 28.2 22.80.032a Vocalizations 64.9 23.4 43.3 26.7 46.1 38.80.082 Disruptive 15.7 27.3 14.0 24.0 12.7 28.40.801
Efficacy of methylphenidate among preschool children
229
Toy changes 10.9 9.9 10.1 9.4 7.1 6.00.240 Epoch 1 5.6 6.8 5.6 7.5 3.3 3.70.311 Epoch 2 4.5 3.7 3.7 2.8 3.1 2.90.476 Epoch 3 0.8 1.1 0.8 0.6 0.7 0.70.914 Compliance task Comply 10 seconds 18.0 14.8 42.1 26.5 44.0 24.60.010a Disruptive 34.5 29.8 25.0 26.3 18.0 34.40.009a Prompts 60.4 16.4 53.4 18.4 52.2 21.10.245 Praise 6.5 4.6 9.2 6.5 9.3 5.50.319 5.2 7.0 2.4 4.1 6.3 10.00.340 Physical guidance Clean-up task % Clean up 24.7 33.0 31.5 31.8 27.1 26.40.556 Disruptive 26.3 34.7 21.1 26.2 32.1 33.10.698 Vocalizations 70.2 14.3 54.5 29.5 67.0 23.30.031a Praise 5.6 8.8 3.2 4.3 7.1 10.70.454 7.5 15.3 4.4 8.1 4.2 12.30.121 Physical guidance a Although the omnibus test was statistically significant, none of the pairwise post hoc tests reached significance. * 0.6 mg/kg>placebo; p<.05; ** 0.6 mg/kg>placebo; p<.005. On the laboratory waiting task, little change was noted on subjects’ on-task or disruptive behavior across drug conditions. The one exception was out-ofseat behavior, which decreased considerably with the introduction of MPH, approaching significance (p<0.057). Minimal improvement in behavior was also noted on the resistance to temptation task, with the exception of a significant decrease in percentage of intervals touching toys. However, although the omnibus test showed a significant decrease for this variable, the pairwise post hoc test results were not significant. During free play we found significant decreases in level of play intensity and movement with the use of medication. Whereas significant improvement between placebo and 0.6 mg/kg dose was noted for the intensity measure, the pairwise post hoc tests did not show a significant decrease in movement. Changes in vocalizations approached significance (p<0.082) and were in the expected direction. Finally, during the compliance task and clean-up, significant decreases were noted in subject vocalizations and disruptive behavior and rates of compliance increased under the MPH conditions. However, none of the pairwise post hoc results were significant. Table 11.2 describes any adverse side effects reported among the subjects as well as posttrial dose recommendations. Five children exhibited significant adverse drug side effects (primarily social withdrawal) during one of the two MPH doses. Subject 11 experienced increased crying, irritability, and social withdrawal during the lower, 0.3 mg/kg dose, and the medication trial was discontinued (the 0.6 mg/kg dose was not given, but the subject was subsequently tried on an open 0.15 mg/kg dose with minimal improvement). A second child (subject 9) experienced severe social withdrawal at the 0.6 mg/kg dose, and medication was discontinued. The remaining three subjects (subjects 1,
Annual progress in child psychiatry and child development 2000–2001
230
6, and 7) also reported significant levels of social withdrawal at the 0.6 mg/kg dose but were able to remain on the dose for the entire 7 days. Problems with social withdrawal were not reported during the 0.3 mg/kg dose (with the exception of subject 11).
TABLE 11.2 Adverse Side Effects and Dose Recommendations
SubjectReported Side Effects Dose Recommendations 1 Good responder at both doses. 0.3 mg/kg: Patient reported increased tearfulness Recommended 0.3 mg/kg dose owing to fewer adverse side effects. 0.6 mg/kg: Patient reported moderate levels of social withdrawal, dullness, staring, and daydreaming. Teacher reported extreme social withdrawal. 2 Minimal to mild responder at both 0.3 mg/kg: Slight increase in reports of staring, social doses. Recommended other stimulants withdrawal, nervous movements; or extended 0.3 mg/kg trial to moderate increase in dullness. determine whether side effects 0.6 mg/kg: No reported concerns. decreased over time. No reported concerns. Best response at 0.6 mg/kg dose. 3 4 0.3 mg/kg: No reported concerns. Best response at 0.3 mg/kg dose. 0.6 mg/kg: Teacher reported increased irritability and crabbiness. 5 No reported concerns. Best response at 0.6 mg/kg dose. Medication not recommended. 6 0.3 mg/kg and 0.6 mg/kg: Significant social withdrawal. 7 0.3 mg/kg: No reported concerns. Responded to both doses. Recommended 0.3 mg/kg dose owing 0.6 mg/kg: Teacher reported significant social withdrawal. to fewer adverse side effects. No reported concerns. Best response at 0.6 mg/kg. 8 9 Recommended 0.3 mg/kg dose or lower 0.3 mg/kg: Mild increase in staring, daydreaming, social owing to mild side effects. withdrawal, and dullness. 0.6 mg/kg: Teacher reported significant social withdrawal, dullness, and irritability. Dose ended early because of concerns. No reported concerns. Best response at 0.6 mg/kg (family 10 chose to postpone medication decision). 11 0.3 mg/kg: Significant increase in Methylphenidate not recommended. irritability, tearfulness, whining, Might consider trial with alternative an anxiety. stimulant medication. 0.6 mg/kg: Not conducted.
Efficacy of methylphenidate among preschool children
231
Table 11.3 presents the mean severity rating (using a 0- to 6-point scale) and number of subjects (frequency) experiencing adverse side effects based on teacher data. Severity ratings and frequency decreased for those “side effects” that are typically associated with ADHD (restlessness, excessive talking). Conversely, severity ratings and frequency increased for the variables dull, social withdrawal, poor appetite, anxiety, and drowsiness.
TABLE 11.3 Mean Severity Rating and Number of Subjects Experiencing Adverse Side Effects
Side Effect Repetitive tongue movements Motor or vocal twitches Nervous movements Tearful, prone to crying Dull, not alert Sad, unhappy, depressed Staring, daydreaming Social withdrawal, talks less Irritable Poor appetite Dizzy, balance unstable Anxiety Headaches Stomach aches, nausea Restless, high activity level Crabby, touchy, whiny Excessive talking Drowsiness Note: Teacher data only.
Placebo Mean (SD) 0.3 (1.0) 0.3 (1.0) 0.4 (1.0) 0.8 (1.5) 0.4 (0.7) 0.2 (0.4) 1.8 (2.1) 0.4 (1.0) 0.6 (1.0) 0.1 (0.3) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 5.0 (1.1) 1.0 (1.4) 2.8 (2.4) 0.0 (0.0)
n 1 1 2 3 3 2 6 2 4 1 0 0 0 0 9 4 7 0
0.3 mg/kg Mean (SD) 0.3 (0.7) 0.3 (0.7) 0.6 (1.1) 0.7 (1.6) 1.5 (1.7) 0.5 (1–6) 2.0 (1.6) 1.3 (1.9) 0.9 (1.5) 1.9 (2.4) 0.2 (0.7) 0.1 (0.3) 0.1 (0.3) 0.1 (0.3) 3.0 (1.6) 1.0 (1.7) 1.0 (1.2) 1.1 (1.8)
n 2 2 3 3 6 1 7 6 3 5 2 1 1 1 6 3 5 3
0.6 mg/kg Mean (SD) 0.4 (1.0) 0.2 (0.6) 0.2 (0.4) 0.1 (0.3) 2.2 (2.0) 0.6 (1.1) 1.7 (2.2) 2.1 (2.4) 1.2 (1.9) 3.2 (2.9) 0.0 (0.0) 0.3 (0.5) 0.0 (0.0) 0.0 (0.0) 1.1 (1.6) 0.7 (1.3) 0.6 (1.1) 0.6 (0.8)
n 2 1 2 1 7 3 5 5 4 6 0 3 0 0 5 4 3 4
Overall, 8 of the 11 children were thought to be drug responders (with a minimum 40% decrease in teacher-rated Conners Hyperactivity Index and/or Behar HyperactiveDistractible subscale between placebo and 1 of the 2 MPH doses), and medication was recommended (6 families chose to follow this recommendation).
DISCUSSION This study found that 8 (73%) of 11 preschool children with developmental disabilities and ADHD had a positive response to stimulant medication. This finding is somewhat above the positive response rates of school-age children with ADHD and mental retardation but consistent with that of typically developing school-age children with ADHD (around 75% to 80%). Those few studies available on MPH efficacy in typically
Annual progress in child psychiatry and child development 2000–2001
232
developing preschool children with ADHD also suggest good response rates among 5year-olds, but poorer response in children younger than 5 years of age. Few adverse side effects have been reported in this group of children. Conversely, adverse side effects were reported in up to 45% of the present sample, a rate considerably greater than that reported among school-age children with developmental disabilities (see Handen et al., 1991) as well as typically developing preschool children. The most commonly reported side effect was social withdrawal, especially at the higher, 0.6 mg/kg dose. The relatively small size of the study cohort and large variance for many dependent measures made it difficult for a number of variables to reach significance. Despite such small numbers, global measures of activity level and inattention consistently evidenced significant decreases when we compared drug conditions to placebo. Conversely, variables measuring conduct problems showed nonsignificant decreases during the drug conditions. This may be because initial conduct problems measures were relatively low at baseline. Laboratory measures were also less robust, but still evidenced a number of significant changes during drug conditions as well as a variety of changes that approached significance. The most apparent gains in behavior with medication were observed during free play and the compliance task. Subjects were found to be less active, vocal, and intense in their play as well as more compliant and less disruptive when parents made requests during the medication conditions. This finding suggests that medication-based changes in behavior among preschool children with developmental disabilities and ADHD may be noted during office visits with minimal additional structure or preparation. Simply observing the child at play may provide evidence of changes in the intensity of play, movement, and vocalizations. Finally, requesting that a parent ask a child to comply with a few simple requests may also show important changes in behavior, especially in terms of compliance and negative behavior exhibited by the child, to comply with a few simple requests may also show important changes in behavior, especially in terms of compliance and negative behavior exhibited by the child. It is interesting that with increased compliance there was minimal change in both the rate of physical assistance and prompting provided by parents. The level of praise offered by parents also appeared to change little, despite an increase in compliance and overall improvement in behavior. This is an important area for intervention. Research with typically developing children has found that as a child’s behavior improves with stimulant medication, adult behavior becomes more positive (Barkley et al., 1984; Whalen et al., 1980). This was not observed in the present study. Although it is not clear why this might be the case, clinicians working with this population must ensure that caretakers are trained to increase their positive feedback to children when they are compliant and well-behaved. Finally, the higher, 0.6 mg/kg drug dose was found to be superior for the majority of variables for which statistically significant findings were noted. In most cases, a change (although not statistically significant) in behavior in the expected direction between placebo and the 0.3 mg/kg dose was also noted. For this population, it may be that the clinician should expect to observe initial improvement at lower dose levels but may find even greater improvement as the dose is titrated toward a 0.6 mg/kg level. However, adverse side effects, such as social withdrawal and irritability, also seem to occur most often at higher MPH doses. As with prior work in this area (Handen et al., 1991), this
Efficacy of methylphenidate among preschool children
233
study’s results suggest that children with developmental disabilities may be at greater risk of developing significant adverse side effects. Clinical Implications Study results suggest that preschool children with developmental disabilities and ADHD respond to MPH at rates similar to those of typically developing children. However, clinicians should be vigilant for significant adverse side effects such as social withdrawal and irritability, especially at the higher, 0.6 mg/kg dose. It is possible that this increased social withdrawal accounts for the improved ratings of ADHD behaviors (e.g., lower activity level, less fidgetiness) in some cases. Therefore, it is important that clinicians examine both behavior rating and adverse side effects scales when making MPH dose decisions. Study results also suggest that clinicians may readily note MPH effects (and side effects) during standard office visits via observations of play and/or motherchild interactions. Changes in behaviors such as motor movements, vocaliza-tions, and compliance to maternal requests as well as side effects such as social withdrawal and irritability may be apparent during clinic visits. Clinicians may also want to work with parents on increasing their level of positive interactions with their children, as this study’s findings suggest that although medication appears to improve overall child behavior, parent-child interactions may actually decrease. Future Research There remain a number of issues for future research in this area. First, this work requires replication with a larger cohort of subjects. Second, potential predictors of both positive MPH response as well as adverse side effects should be examined. Finally, research on the possible additive effects of nonpharmacological treatment and MPH is warranted. It is possible that similar rates of efficacy can be obtained with the combination of lower MPH doses and behavioral intervention, thereby decreasing the potential for adverse side effects.
ACKNOWLEDGMENTS This research was supported, in part, by a grant to the first author from the Fanny Pushin Rosenberg Research Foundation. The authors thank Joseph Mazzotta, R.Ph., of the Children’s Hospital of Pittsburgh Pharmacy Department and Rea Ann Maxwell, R.Ph., of the Western Psychiatric Institute and Clinic Pharmacy Department for their assistance in preparing the methylphenidate doses and David Macarchick, B.S., for his assistance in conducting the statistical analysis.
Annual progress in child psychiatry and child development 2000–2001
234
REFERENCES Aman MG, Kern RA, McGhee DE, Arnold E (1993). Fenfluramine and methylphenidate in children with mental retardation and ADHD: clinical and side effects. J Am Acad Child Adolesc Psychiatry 32:851–859. Aman MG, Marks RE, Turbott SH, Wilsher CP, Merry SN (1991). The clinical effects of methylphenidate and thioridazine in intellectually subaverage children. J Am Acad Child Adolesc Psychiatry 30:246–256. American Psychiatric Association (1987), Diagnostic and Statistical Manual of Mental Disorders, 3rd edition-revised. Washington, DC: Author. Barkley RA (1988). The effects of methylphenidate on the interactions of preschool ADHD children with their mothers. J Am Acad Child Adolesc Psychiatry 27:336–341. Barkley RA (1990). Attention deficit hyperactivity disorder: A handbook for diagnosis and treatment. New York: Guilford. Barkley RA (1998). Attention-deficit/hyperactivity disorder. In: Treatment of Childhood Disorders, 2nd ed, Mash EJ, Barkley RA, eds. New York: Guilford, pp 55–110. Barkley RA, DuPaul G, Costello A (1993). Stimulants. In: Practitioner’s Guide to Psychoactive Drugs for Children and Adolescents, Werry JS, Aman MG, eds. New York: Plenum Medical Book Company, pp 205–237. Barkley RA, Karlsson J, Strzelecki E, Murphy JV (1984). Effects of age and Ritalin dosage on mother-child interactions of hyperactive children. J Consult Clin Psychol 52:750–758. Behar L, Stringfield S (1974). A behavior rating scale for the preschool child. Dev Psychol 10:601–610. Collier RO, Baker F, Mandville G, Hoyes T (1972). Estimates of test size for several test procedures based on conventional variance ratios in the repeated measures design. Psychometrika 32:339–353. Conners CK (1975). Controlled trial of methylphenidate in preschool children with minimal brain dysfunction. Int J Ment Health 4:61–74. Davidson ML (1972). Univariate versus multivariate tests in repeated measures experiments. Psychol Bull 77:446–452. Davy T, Rodgers CL (1989). Stimulant medication and short attention span: a clinical approach. J Dev Behav Pediatr 10:313–318. Gadow KD (1977). Psychotropic and Antiepileptic Drug Treatment With Children in Early Childhood Special Education. Champaign: University of Illinois, Institute for Child Behavior and Development (ERIC Document Reproduction Service Number ED 162 294). Gadow KD (1981). Prevalence of drug treatment for hyperactivity and other child behavior disorders. In: Psychosocial Aspects of Drug Treatment for Hyperactivity, Gadow KD, Loney J, eds. Boulder, CO: Westview Press, pp 13–76. Gadow KD, Pomeroy JC (1990). A controlled case study of methylphenidate and fenfluramine in a young mentally retarded, hyperactive child. Aust NZ J Dev Disabil 16:323–344.
Efficacy of methylphenidate among preschool children
235
Goyette CH, Conners CK, Ulrich RF (1978). Normative data on revised Conners Parent and Teacher Rating Scales. J Abnorm Child Psychol 6:221–236. Handen BL, Breaux AM, Gosling A, Ploof DL, Feldman H (1990). Efficacy of methylphenidate among mentally retarded children with attention deficit hyperactivity disorder. Pediatrics 86:922–930. Handen BL, Breaux AM, Janosky J, McAuliffe S, Feldman H, Gosling A (1992a). Effects and noneffects of methylphenidate in children with mental retardation and ADHD. J Am Acad Child Adolesc Psychiatry 31:455–461. Handen BL, Feldman H, Gosling A, Breaux AM, McAuliffe S (1991). Adverse side effects of methylphenidate among mentally retarded children with ADHD. J Am Acad Child Adolesc Psychiatry 30:241–245. Handen BL, McAuliffe S, Janosky J, Breaux AM, Feldman H (1995). Methylphenidate in children with mental retardation and ADHD: effects on independent play and academic functioning. J Dev Phys Disabil 7:91–103. Handen BL, Parrish JM, McClung TJ, Kerwin ME, Evans LD (1992b). Using guided compliance versus time out to promote child compliance: a preliminary comparative analysis in an analogue context. Res Dev Disabil 13:159–170. Helsel WJ, Hersen M, Lubetsky MJ, Fultz SA, Harlovic CH (1989). Stimulant drug treatment of four multihandicapped children using a randomized single-case design. J Multihandicapped Person 2:139–154. Kolmen BK, Feldman H, Handen BL, Janosky JE (1995). Naltrexone in young autistic children: a double-blind placebo-controlled crossover study. J Am Acad Child Adolesc Psychiatry 34:223–231. Marascuilo LA, Levin JR (1983). Multivariate Statistics in the Social Sciences: A Researcher’s Guide. Monterey, CA: Brooks/Cole Publishing Company. Physicians’ Desk Reference (1988). Oradell, NJ: Medical Economics Company. Physicians’Desk Reference (1996). Oradell, NJ: Medical Economics Company. Rosenberg MS (1987). Psychopharmacological interventions with young hyperactive children. Top Early Child Spec Educ 6:62–74. Schell RM, Pelham WE, Bender ME, Andree JA, Law R, Robbins F (1986). The concurrent assessment of behavioral and psychostimulant interventions: a controlled case study. Behav Assess 8:373–384. Schleifer M, Weiss G, Cohen N, Elman M, Cvejic H, Kruger E (1975). Hyperactivity in preschoolers and the effect of methylphenidate. Am J Orthopsychiatry 45:38–50. Shafto F, Sulzbacher S (1977). Comparing treatment tactics with a hyperactive preschool child: stimulant medication and programmed teacher intervention. J Appl Behav Anal 10:13–20. Whalen CK, Henker B, Dotemoto S (1980). Methylphenidate and hyperactivity: effects on teacher behaviors. Science 208:1280–1282. Wilens TE, Biederman J (1992). The stimulants. Psychiatr Clin North Am 15:191–222.
PART III: ATTENTION DEFICIT-HYPERACTIVITY DISORDERS
12 ADHD in Girls: Clinical Comparability of a Research Sample Wendy S.Sharp, James M.Walter, Wendy L.Marsh, Gail F.Ritchie, Susan D.Hamburger, and F.Xavier Castellanos
Objective: The investigation of attention deficit-hyperactivity disorder (ADHD) in girls raises complex questions of referral bias and selection criteria. The authors sought to determine whether they could recruit a research sample of comparably affected girls using a combination of sex-independent diagnostic criteria and sex-normed cutoffs on teacher ratings. They also report on the largest placebocontrolled crossover comparison of methylphenidate and dextroamphetamine in girls with ADHD. Method: Subjects were 42 girls with DSM-III-R/DSM-IV ADHD (combined type) contrasted to 56 previously studied boys with ADHD on comorbid diagnoses, behavioral ratings, psychological measures, psychiatric family history, and stimulant drug response. Results: Girls with ADHD were statistically indistinguishable from comparison boys on nearly all measures. Girls exhibited robust beneficial effects on both stimulants, with nearly all (95%) responding favorably to one or both drugs in this short-term trial. Dextroamphetamine produced significantly greater weight loss than methylphenidate. Conclusions: This highly selected group of ADHD girls was strikingly comparable with comparison boys on a wide range of measures. The results confirm that girls with ADHD do not differ from boys in response to methylphenidate and dextroamphetamine and that both stimulants should be tried when response to the first is not optimal. (J. Am. Acad. Child Adolesc. Psychiatry, 1999, 38(1):40–47.) Key Words: attention deficithyperactivity disorder, sex differences, randomized clinical trials, methylphenidate, dextroamphetamine.
INTRODUCTION The literature on attention deficit-hyperactivity disorder (ADHD) in girls is scant and inconsistent. Almost all research on ADHD has focused exclusively on boys. This bias reflects male-female ratios in referred samples ranging from 4:1 to 9:1 (American
ADHD in girls
237
Psychiatric Association, 1994) and the perceived need for homogeneity in research samples. However, community-based studies have found male-female sex ratios as low as 2.1:1 (Szatmari, 1992; Taylor et al., 1998), confirming that girls with ADHD have been neglected by clinicians and researchers (Berry et al., 1985). The discrepancy between clinic and community rates of ADHD in boys and girls and the questions it raises were the focus of a National Institute of Mental Health (NIMH) Conference on Sex Differences in ADHD (Arnold, 1996b). The participants noted the existence of substantial evidence of normative sex differences that influence the manifestations of ADHD, so that the issue of selecting comparable sex-matched subjects for study is not trivial. For example, if identical criteria are used for both boys and girls with ADHD, when the normative populations differ in symptom distribution, then the few girls who meet selection criteria would be expected to exhibit markedly greater severity relative to their same-sex controls. On the other hand, using completely distinct criteria for both sexes, e.g., exceeding the 95th percentile for that sex on all measures, might include girls who are not comparably impaired by their symptoms. When we embarked on the study of ADHD in girls in 1993, we decided to combine the two approaches. That is, we required that all subjects meet full criteria for ADHD (initially DSM-III-R, later combined type DSM-IV). We also required that teachers’ hyperactivity ratings exceed the 95th percentile, but set distinct sex-normed cutoffs for boys and girls. We hypothesized that this would allow us to recruit a sample of girls with ADHD of comparable severity to that of previously studied boys. We are now able to examine the results of our recruitment and screening efforts as we prepare data analyses of brain anatomy in girls with ADHD. There has also been a paucity of randomized, controlled medication trials in girls with ADHD. Two small studies found no differences between boys and girls in response to methylphenidate (Barkley, 1989; Pelham et al., 1989), but there have been no controlled studies of the efficacy of dextroamphetamine in girls. We now report on the largest sample of girls with ADHD to undergo a placebo-controlled crossover comparison of methylphenidate (MPH) and dextroamphetamine (DEX).
METHOD Subjects Girls with a history of severe hyperactivity, impulsivity, and inattentiveness that interfered with home and school functioning were recruited for controlled stimulant trials and anatomic brain imaging from local schools and health care providers beginning in 1993. Structured telephone screenings were conducted by research social workers (G.F.R., W.S.S.) to determine that symptoms of ADHD were present in at least two settings and that Conners Hyperactivity factor scores from their home teacher were at least 2 SDs greater than age and sex norms (≥1.0 for girls versus ≥1.8 for boys, scored from 0 to 3; Werry et al., 1975). Exclusion criteria were Full Scale IQ less than 80 on the WISC-R (Wechsler, 1974) and chronic medical or neurological diseases, including Tourette disorder and chronic tic disorders.
Annual progress in child psychiatry and child development 2000–2001
238
From approximately 150 initial inquiries, 65 subjects declined to participate (42 after telephone screening and 23 after submitting initial rating scales) for a variety of reasons, including improvement in symptoms, parents’ discomfort with medication, or lack of follow-through. We excluded 25 subjects primarily because their hyperactivity symptoms were not sufficiently severe. Other exclusion reasons included the following: IQ <80 (n=4), other medical conditions (n=5), and medications other than stimulants that could not be discontinued (n=2). Five subjects are not included in the main analyses because they did not also enter the neuroimaging study component. Two other subjects were excluded after completing the study. In one case, we uncovered ongoing sexual abuse that raised questions about the validity of the ADHD diagnosis. In the other, we concluded at the completion of the 3-month day program that impairment was mostly ascribable to severe multiple learning disorders rather than ADHD symptoms. Of the remaining 42 girls, 67% are white, 19% are African American, and 14% are Latina. The most common referral sources were local schools (35%) and physicians (24%). Children were also referred by program alumni (10%), the National Institutes of Health (NIH) listing of clinical studies (10%), friends (7%), and the advocacy organization Children and Adults with Attentional Disorders (2%); 2% learned of our study from media reports. Referral sources were unknown for 10%. Besides the 32 girls who participated in the controlled stimulant trial described below, a second group (n=10) participated in a pilot study comparing sustained-release DEX, placebo, and the mixed amphetamine compound, Adderall® (unpublished, 1998). Recruitment strategies, the location of the study, the personnel, curriculum, scheduling, and the duration of the studies were identical. The two groups of girls did not differ significantly on any measure (data available on request). Comparison Group The 56 comparison subjects were all the subjects included in a prior publication (Castellanos et al., 1996b) except for one boy with chronic motor tics. All but one of the boys had also participated in controlled stimulant trials in the same research day program (Castellanos et al., 1996a; Elia et al., 1991). Seventy-three percent of the boys are white, 21% are African American, 4% are Latino, and 2% are Asian American. Measures Final DSM-IV diagnoses were obtained by a child and adolescent psychiatrist combining information from clinical interviews, staff observations, teacher ratings, and parent structured interview with the Diagnostic Interview for Children and Adolescents-Parent Version (Herjanic & Campbell, 1977). Psychiatric diagnoses of day program probands’ biological parents were obtained by unblinded in-person interviews using the Schedule for Affective Disorders and Schizophrenia (Spitzer & Endicott, 1982). We obtained the Wender Utah Rating Scale (WURS) (Ward et al., 1993) from available biological parents to obtain a dimensional measure of childhood ADHD symptoms. Provisional categorical classification of parents into ADHD or non-ADHD groups was performed by using 95th percentile cutoffs for each sex (32 or greater for females, 40 or greater for males) (Ward
ADHD in girls
239
et al., 1993). Information on ADHD status of siblings was gathered via genograms by requesting that parents categorize the ADHD status of all full siblings aged 7 or older as absent, probable, or definite. Siblings who had been given a medical diagnosis of ADHD and were receiving ongoing stimulant treatment were classified as having definite ADHD. Psychoeducational evaluation consisted of the WISC-R (Wechsler, 1974) and Woodcock-Johnson Achievement Battery (Woodcock & Johnson, 1977), performed by a psychologist (B.B.K.). Ten of the most recently recruited girls were assessed with the updated versions (WISC-III, Wechsler, 1991; Woodcock-Johnson Revised, Woodcock & Johnson, 1989). Children were classified as having reading disorder if their IQ-Reading discrepancy z score exceeded 1.65 (Frick et al., 1991). We obtained Conners Hyperactivity and Conduct factors (range 0 to 3), the Child Behavior Checklist (CBCL), and the Teacher’s Report Form (TRF) (Achenbach & Edelbrock, 1983) from parents and from the children’s home teacher. Overall impairment was quantified using the Children’s Global Assessment Scale (C-GAS) (range 0 to 100) (Shaffer et al., 1983) and the Clinical Global Impressions scale for Severity of Illness (CGISI, range 1 to 7) (Clinical Global Impressions, 1985). Commission and omission errors on the continuous performance test (CPT) (Rosvold et al., 1956) were obtained during drug-free and/or during placebo phase. Most of the boys had been administered CPT during baseline and drug double-blind periods, thus including placebo. Because of a change in the CPT testing schedule, baseline-only data were available for 21 girls (60%) and placebo-only data for 14 girls (40%). For comparison boys, we randomly selected data from baseline for 32 boys (60%) and for the placebo period for 21 boys (40%) to control for practice effects. Controlled Trial of Methylphenidate, Dextroamphetamine, and Placebo Children attended our accredited NIMH school 5 days a week for 3 months with academic instruction in the morning and recreation therapy activities in the afternoon. MPH, DEX, and placebo were packaged in identical capsules by the NIH Pharmacy and administered by NIH nurses in double-blind, randomized order at breakfast and lunch 5 days per week (and by parents on weekends) after a 3-week medication-free baseline. Individual doses were packaged in coded blister packs. Weekend medication compliance was confirmed by weekly telephone contacts. Individual drug dosages were selected for each subject prior to study entry based on body weight and medication history; all subjects underwent stepwise increases in their stimulant dose each week. Each doubleblind phase lasted 3 weeks. Conventional doses were used in girls, with daily doses of MPH ranging from 10 mg to 70 mg/day and of DEX from 5 mg to 30 mg/day in two divided doses. Mean stimulant doses were 0.45, 0.85, and 1.28 mg/kg per dose for MPH, and 0.23, 0.43, and 0.64 mg/kg per dose for DEX for weeks 1, 2, and 3, respectively. The comparison boys had received a higher range of daily doses (MPH 25 to 90 mg, DEX 10 to 45 mg), which had been planned to minimize stimulant non-response due to potentially inadequate dosing (Elia et al., 1991). Weekly outcome measures were parent and NIMH teacher Conners ratings of hyperactivity and conduct, and physician-rated global severity (CGI-SI, CGAS), global
Annual progress in child psychiatry and child development 2000–2001
240
improvement (Clinical Global Impressions-Global Improvement [CGI-GI]) (citations noted), stimulant-related adverse effects (Guy, 1976), and body weight. The studies were approved by the NIMH Institutional Review Board, and signed consent and assent were obtained from all parents and children, respectively. Statistical Analysis Analyses were performed using SAS for Windows, version 6 (SAS Institute, 1996). Oneway ANOVA and t-tests were used to compare baseline measures, including socioeconomic status (Hollingshead, 1975) between the 32 girls in the MPH/DEX trial and the 10 girls who took part in the pilot amphetamine trial. Because they did not differ significantly on any baseline measure, the two subsamples were combined and compared with previously studied boys. Repeated-measure ANOVA was used to examine drug and dose effects. Significant ANOVA results were further explored with preplanned Bonferroni ttests. For example, significant drug by dose interactions were explored by testing the effects of the highest MPH dose versus the effects of the highest DEX dose. Carryover effects were tested by comparing the teachers’ ratings of hyperactivity during the first week across the six different randomization schedules using ANOVA for MPH, DEX, and placebo. Drug response was defined as a rating of “very much” or “much improved” on the CGI-GI. Dimensional measures of drug response were calculated as the difference between the average of each subject’s three weekly teachers’ hyperactivity ratings during the placebo phase and ratings for their best week during the medication phases (Rapport et al., 1986). Stepwise multiple regression was used to determine significant moderators of drug response. Because of extreme outliers, the median test was used to compare girls with boys on CPT commission and omission errors. Chi-square (or Fisher exact test when appropriate) was used to analyze psychiatric diagnoses in subjects and parents of subjects. Parents’ psychiatric diagnoses were grouped into affective (major depression, dysthymia, and bipolar disorder), anxiety (panic disorder, generalized anxiety disorder, obsessivecompulsive disorder, and phobic disorder), and substance abuse disorders (alcohol abuse and substance abuse). Missing weekly body weight data (9%) were calculated by interpolation. Other missing data (1%) were replaced by cell means. All tests were 2tailed with α=0.05.
RESULTS Table 1 shows characteristics of the study subjects and the male comparison group. Female subjects and their male cohorts did not differ significantly on demographic or psychoeducational measures, with four exceptions. Girls had significantly lower Woodcock-Johnson Reading Standard Scores than boys (p =0.04). Psychoeducational scores were unavailable for 1 boy and for 2 girls who were tested elsewhere too recently to allow valid retesting. Achievement scores of 13 boys who had been given a different test were not included.
ADHD in girls
241
TABLE 12.1 Means and Standard Deviations of Clinical Characteristics of Girls and Comparison Boys With ADHD
Girls With ADHD Comparison Boys t df p (n=42) (n=56) Age (yr) 8.9±1.7 9.3±1.7 −1.21 96 0.23 Age range (yr) 6.2–12.7 6.0–12.5 3.6±2.1 2.8±1.9 −1.71 74 0.09 Age at onset (yr) SES 48.0±25.8 52.4±26.9 −0.81 94 0.42 WISC-R Full Scale IQ 105.2±12.8 109.3±17.7 −1.10 82 0.28 WISC-R Verbal IQ 105.6±14.7 109.7±19.8 −0.98 82 0.33 WISC-R Performance 104.0±12.9 107.0±15.8 −0.88 82 0.38 IQ Woodcock- Johnson 95.6±14.3 103.8±16.5 −2.10 67 0.04* Reading Standard Score Woodcock-Johnson 96.6±14.5 102.7±19.5 −1.40 67 0.17 Math Standard Score C-GAS 44.6±4.8 45.3±5.9 −0.66 94 0.51 CGI-SI 5.0±0.9 4.4±0.7 4.10 89 0.0001* Teacher Conners 2.0±0.6 2.4±0.5 −3.17 90 0.002* Hyperactivity factor Conduct factor 0.9±0.7 1.1±0.5 −1.85 63 0.07 Parent Conners Hyperactivity factor 2.5±0.5 2.4±0.5 0.96 89 0.34 Conduct factor 1.4±0.6 1.4±0.6 −0.18 89 0.86 Child Behavior Checklist Attention Problems 76.0±7.4 72.4±8.2 2.07 83 0.04* 70.7±8.8 68.0±9.0 1.39 83 0.17 Externalizing a Behaviors 63.6±7.8 63.7±9.7 −0.08 83 0.94 Internalizing Behaviorsb Total Behavior 71.0±7.0 69.9±7.4 0.70 83 0.49 Problems Teacher’s Report Form Attention Problems 70.3±8.0 70.0±9.5 0.14 76 0.89 69.7±8.5 68.4±6.6 0.74 76 0.46 Externalizing Behaviorsa 61.0±8.1 63.3±10.1 −1.08 76 0.28 Internalizing b Behaviors Total Behavior 69.3±6.3 70.2±6.9 −0.63 76 0.53 Problems Note: SES=socioeconomic status; C-GAS=Children’s Global Assessment Scale;
Annual progress in child psychiatry and child development 2000–2001
242
CGI-SI= Clinical Global Impressions, Severity of Illness. a Attention Problems, Delinquent Behaviors, Aggressive Behaviors. b Withdrawn, Somatic Complaints, Anxious/Depressed, Social Problems, Thought Problems. *p<.05. Conners teacher, but not parent, ratings of hyperactivity were significantly higher for boys than for girls (p=0.002), which was not surprising, because the normative sexappropriate cutoffs we used were higher for boys (1.8 versus 1.0 for girls) (Werry et al., 1975). For the CBCL and TRF factors relating to internalizing (Withdrawn, Somatic Complaints, Anxious/Depressed, Social Problems, and Thought Problems) and externalizing behaviors (Attention Problems, Delinquent Behaviors, and Aggressive Behaviors), parent data were available for all 42 girls and 43 boys, and teacher data for 40 girls and 38 boys. No significant differences were found for girls with ADHD and comparison boys for either internalizing or externalizing symptoms except that parents rated Attention Problems as significantly more severe for girls (p=0.04). Globally, girls were rated as significantly more impaired (p=0.0001) on the categorical CGI-SI, but the sexes did not differ on the continuous C-GAS measure of global functioning. Comparison boys made more omission and commission errors than girls, although these differences did not reach significance (median test, p=0.19, p =0.38, respectively). Comorbidity Girls and boys with ADHD were similar in comorbidity, whether defined in general (ADHD plus at least one other diagnosis) (girls 69%, boys 71%) or by individual analyses for all diagnoses that were present in either group. Those rates were oppositional definat disorder (girls 50%, boys 33%, p=0.09), conduct disorder (girls 2%, boys 7%), major depression (girls 7%, boys 0%, p= 0.08), separation anxiety (girls 2%, boys 0%), specific phobias (girls 7%, boys 0%, p=0.08), trichotillomania (girls 0%, boys 2%), tic disorders not otherwise specified (girls 2%, boys 13%, p=0.13), enuresis (girls 12%, boys 18%), and reading disorder (girls 8%, boys 5%). Parents’ Diagnoses There were no significant differences between parents of day program girls and parents of comparison boys on affective disorders (30% and 39%, respectively), anxiety disorders (25% and 14%, respectively, p=0.08), or substance abuse disorders (19% and 16%, respectively). Twenty parents of girls (24%) and 30 parents of boys (24%) did not meet criteria for any psychiatric diagnosis. Parental self-report of childhood ADHD symptoms, as quantified by WURS scores, did not differ significantly in dimensional analysis. Group means were 25.0±18.1 and 33.7±17.6 for fathers and mothers of girls, respectively; and 29.0±15.1 and 26.1±20.4 for fathers and mothers of boys, respectively. When we used categorical cutoffs (≥40 for fathers and ≥32 for mothers), a larger proportion of the parents of girls than of boys had
ADHD in girls
243
scores in the ADHD clinical range (47% versus 24%, p=ø.007). This difference was found only in mothers of girls compared with mothers of boys (62% versus 31%, p=0.008); the difference between fathers of girls and fathers of boys was not significant (30% versus 21%, respectively). ADHD in Siblings A significantly higher proportion of full siblings (aged 7 or older) of girls with ADHD were categorized as having definite or probable ADHD compared with siblings of comparison boys (50% versus 16%, p=0.004), regardless of whether the sibling was male or female. Controlled Comparison of Methylphenidate, Dextroamphetamine, and Placebo All subjects completed the trial with the exception of one 6-year-old for whom the placebo phase was blindly truncated to 2 weeks because of her severe physical impulsivity, without informing staff, parents, or the child. In this case, last observations were carried forward. Remarkably, none of the 180 possible pairwise comparisons (4 measures× 3 drug phases×15 pairs) yielded significantly different results on carryover analysis. Individual weekly ratings demonstrated highly significant main effects of drug (F>58.22, p<0.0001) and of dose (F>15.06, p<0.0001) on all measures demonstrating robust dose-related stimulant effects relative to placebo. Absence of “dose-related” change on placebo was highlighted by more moderate, although still significant, drug by dose interactions for all four weekly measures: CGI-SI (F=2.56, p=0.04), C-GAS (F=6.76, p=0.0001), and Conners teacher and parent Hyperactivity factor (F=9.21, p=0.0001; F= 4.08, p=0.004, respectively). Global improvement (CGI-GI) is illustrated in Figure 12.1. Five girls (16%) were judged to have improved substantially on placebo. Twenty-two girls (69%) improved substantially on both MPH and DEX. Nine of the remaining 10 responded to one stimulant but not the other (four to MPH, five to DEX). Thus, the response rate to either MPH or DEX for study completers was 97%. We found the same high total rate of response (37/39=95%) when we included all initially enrolled subjects. For the comparison boys, 69% responded to MPH and 72% to DEX, with a response rate of 87% to one or the other stimulant. Mean beneficial and adverse effects of DEX and MPH were nearly identical for all ratings, including ratings of appetite problems. However, objectively verified significant decreases in body weight (drug main effect, F=10.27, p= 0.0002) were significantly greater for DEX (mean change −1.1±1.0 kg from baseline, p=0.01) than for MPH (−0.4±1.1 kg, not significant).
Annual progress in child psychiatry and child development 2000–2001
244
Figure 12.1. Percentages of 32 girls and 45 boys with attention deficit-hyperactivity disorder who had double-blind Clinical Global Impression-Global Improvement ratings of “very much improved” or “much improved.” DEX=dextroamphetamine; MPH= methylphenidate. The only significant predictor of improvement on teacher ratings of hyperactivity was baseline severity (R2=0.54, p<.0001). Thirty-one of the 32 female subjects were prescribed a stimulant at discharge: 47% received MPH (29.0 mg/day±15.1) and 50% received DEX (18.1 mg/day±6.2). Stimulant medications were not recommended at discharge for one subject because neither medication substantially improved her ADHD behaviors beyond placebo. One child who was classified as a nonresponder on all three phases while in the day program exhibited moderately improved behaviors on DEX at home and was discharged on DEX 7.5 mg b.i.d. Of the comparison boys, 51% had been discharged on MPH (46.1 mg/day±20.9), 45% on DEX (27.1 mg/day± 8.9), and two children with the recommendation of either stimulant. Ten boys were in a pemoline/placebo study, and their results are not included.
DISCUSSION In this referred and highly selected sample (<30% of initial inquiries), girls with DSM-IV combined type ADHD were strikingly similar to the boys with ADHD we previously studied. On the other hand, with the exception of the Conners teachers’ ratings of hyperactivity, when there were significant differences, they were in the direction of greater severity for girls than for boys. Thus, girls had significantly lower reading scores than did boys, although the prevalence of reading disorder was low in both groups. Of the CBCL factors, the Attention Problems score is the best positive predictor of ADHD diagnosis (Hudziak, 1997). Parent-rated Attention Problems T scores were significantly higher in the girls than in the boys, but this difference was not supported by the comparable teacher ratings. On physician-rated global impairment, girls had higher
ADHD in girls
245
severity than boys on one scale (CGI-SI) but not on another (C-GAS). Diagnostically, there were no significant differences in comorbidity patterns in probands or in their biological parents, although there were statistical trends toward a higher prevalence of oppositional defiant disorder, major depression, and specific phobia in girls. The overall pattern of comparable impairment in referred girls with ADHD is consistent with a recent meta-analysis (Gaub & Carlson, 1997). The tendency toward somewhat greater severity on some measures also echoes recent findings in an epidemiologically ascertained sample (Heptinstall et al., 1998). Also in support of the hypothesis that girls who are referred represent a more extreme sample than clinic-referred boys was our finding of greater familiality of ADHD for parents (at least for mothers) and siblings of girls with ADHD in comparison with relatives of boys. The largest communitybased twin study found that females with ADHD have a higher frequency of first-degree relatives with ADHD than do ADHD males (Rhee et al., in press). Their analyses were consistent with a multiple threshold model for the sex differences in ADHD, with diagnosed females having a higher threshold. In clinicreferred samples, results differ. Two studies found no significant difference in familiality based on sex (Faraone et al., 1991; Mannuzza & Gittleman, 1984), but another study reported greater familiality in families of female probands (only in families with antisocial disorders) (Faraone et al., 1995). Confirmation of greater familiality in families of female probands would have substantial implications for genetic studies. However, we view our results with caution for three reasons. First, we did not obtain the samples contemporaneously; thus we did not control cohort effects, such as the dramatic increase in the rate of diagnosis and stimulant treatment of ADHD in the 1990s (Safer et al., 1996). Second, parental ADHD status was determined from a single selfreport checklist (WURS) rather than from structured retrospective interviews, and we did not obtain collateral documentation (such as old report cards or grandparent reports). Finally, we did not obtain blinded structured psychiatric interviews for siblings, but rather used a brief genogram interview and/or family medical history to ascertain presence or absence of ADHD. Thus, our primary conclusion is that our sample of girls demonstrated very similar patterns of comorbidity and impairment and identical patterns of drug response. Their neurobiological data should be informative when compared with and contrasted to that of our previously studied boys, particularly because brain structures of interest in ADHD, such as the caudate nucleus (Swanson et al., 1998), are sexually dimorphic in healthy children, with proportionately larger volumes in girls (Giedd et al., 1997). Clinical Implications: Sex Differences in ADHD There is a popular notion that girls with ADHD have primarily attentional difficulties and a later age at onset than boys. In our samples, age at onset did not differ significantly. We also found a higher frequency of oppositional defiant disorder in our sample of girls, possibly because of selection and referral bias. The following vignettes provide a qualitative “flavor” of a representative selection of our subjects. Case 1. A was a 6-year-old who began to take MPH at age 2½ years, when. ADHD was diagnosed. Her family history is negative for ADHD. She was asked to leave a
Annual progress in child psychiatry and child development 2000–2001
246
prekindergarten program because of her “disruptive behavior.” Although academically on grade level, she jumped from task to task, had difficulty focusing, marked furniture with crayons, and crawled and hid under her desk. At home, she was in constant motion; she ran into the street several times without checking for cars, narrowly avoiding serious accidents. During the study, A almost lost transportation privileges because she did not stay seated on the van. Her placebo phase was shortened because of severe impulsivity that endangered her safety. At the time of discharge, A was receiving 10 mg of DEX twice a day. Case 2. B was a 7-year-old who began to take MPH at age 6 years, when ADHD was diagnosed. Her brother and mother also had ADHD. B’s teacher noted that she reacted quickly and impulsively, without thinking about consequences, and that peers did not want to sit near her because she hit or kicked them. Her parents complained that B sat only briefly and often ate dinner swinging her legs or sitting on her knees. B ran away from her parents or hit them when frustrated. During the study, B was usually able to stay in her chair but the chair and desk would gradually move across the classroom as a result of her constant fidgeting. B also had oppositional defiant disorder; at discharge she was prescribed 5 mg of MPH each morning and 2.5 mg at lunch. Case 3. C was an 11-year-old whose ADHD was diagnosed at age 6 years. A brief trial of MPH was discontinued because of maternal concern. C’s father may have had ADHD as a child; otherwise, the family history was reported as negative. Her mother described C as stubborn, impatient, and intrusive. Her performance was below grade level in math and reading, and she could not stay in her seat at school. C was suspended twice for physical aggression. During the study, C threatened students and once shoved a classmate. C denied responsibility when confronted, and she appeared unaware of others’ “personal space.” C’s artwork often displayed poor self-control; she once covered her paper with a thick layer of black chalk and then proceeded to smear black chalk on the bathroom walls. C also had oppositional defiant disorder; at discharge she was prescribed a 15-mg DEX Spansule® each morning and a 5-mg DEX tablet after school. Clinical Implications: Controlled Trial of Methylphenidate and Dextroamphetamine Previous smaller studies of girls with ADHD have found that the response to MPH is comparable in both sexes (Barkley, 1989; Pelham et al., 1989). In this, the largest controlled stimulant trial in girls with ADHD, we replicated the prior observation in boys that MPH and DEX are comparably effective and that the rate of efficacy is even higher when both drugs are considered (Elia et al., 1991). Our findings were strikingly robust, especially considering that we used lower doses and a more conservative definition of drug response than had Elia et al. (1991). It thus supports the recommendation that whenever the response to the first stimulant tested is suboptimal, the alternative stimulant should be tried (Arnold, 1996a). However, our results are derived from a short-term trial in a structured research day program in highly selected subjects, and thus they may not apply to samples with more comorbid disorders (Tannock et al., 1995) or less distinct cases of combined type ADHD (Spencer et al., 1996). Regarding choice of first stimulant, MPH and DEX were indistinguishable on all
ADHD in girls
247
measures of efficacy and adverse effects with one exception. DEX produced a significant mean loss in body weight, whereas MPH did not. Our data thus provide additional support (see also Efron et al., 1997a,b) for the usual clinical practice of beginning most stimulant trials with MPH. Other Limitations As noted, the present sample and the comparison boys were not studied contemporaneously. Nevertheless, the continuity of program, staff, referral sources, and diagnostic instruments appears to have mitigated this potential confound. We also combined subjects from two separate clinical trials, but we found that the two subsamples were statistically indistinguishable, and our conclusions would have been unchanged if we had reduced the sample of girls to 32. Finally, in the interest of continuity, we used older versions of psychoeducational instruments, thus necessitating caution in comparing our specific numerical values to other samples.
ACKNOWLEDGMENTS The authors appreciate the assistance of NIH Day Program teachers Phyllis Siegrist, M.Ed., and Anna Davidson, M.Ed.; recreation therapists Debbie C. Marcus, C.T.R.S., and Robin Greenfield, C.T.R.S.; NIH Clinical Center staff members Kathy Berzin, R.N., and Eddie Frazier, R.N.; consulting psychologist Barbara B.Keller, Ph.D. The authors are grateful for the participation of the children and their families and for the support and leadership of Judith L. Rapoport, M.D.
REFERENCES Achenbach T, Edelbrock C (1983). Manual for the Child Behavior Checklist and Revised Child Behavior Profile. Burlington: University of Vermont Department of Psychiatry. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Arnold LE (1996a). Responders and nonresponders (letter). J Am Acad Child Adolesc Psychiatry 35:1569–1570. Arnold LE (1996b). Sex differences in ADHD: conference summary. J Abnorm Child Psychol 24:555–569. Barkley RA (1989). Hyperactive girls and boys: stimulant drug effects on motherchild interactions. J Child Psychol Psychiatry 30:379–390. Berry CA, Shaywitz SE, Shaywitz BA (1985). Girls with attention deficit disorder: a silent minority? A report on behavioral and cognitive characteristics. Pediatrics 76:801–809. Castellanos FX, Elia J, Kruesi MJP et al. (1996a). Cerebrospinal homovanillic acid predicts behavioral response to stimulants in 45 boys with attentiondeficit/hyperactivity disorder. Neuropsychopharmacology 14:125–137.
Annual progress in child psychiatry and child development 2000–2001
248
Castellanos FX, Giedd JN, Marsh WL et al. (1996b). Quantitative brain magnetic resonance imaging in attention-deficit/hyperactivity disorder. Arch Gen Psychiatry 53:607–616. Clinical Global Impressions (1985). Psychopharmacol Bull 21:839–843. Efron D, Jarman F, Barker M (1997a). Methylphenidate versus dexamphetamine in children with attention deficit hyperactivity disorder: a double-blind, crossover trial. Pediatrics 100:E61-E67. Efron D, Jarman F, Barker M (1997b). Side effects of methylphenidate and dexamphetamine in children with attention deficit hyperactivity disorder: a doubleblind, crossover trial. Pediatrics 100:662–666. Elia J, Borcherding BG, Rapoport JL, Keysor CS (1991). Methylphenidate and dextroamphetamine treatments of hyperactivity; are there true non-responders? Psychiatry Res 36:141–155. Faraone SV, Biederman J, Chen WJ, Milberger S, Warburton R, Tsuang MT (1995). Genetic heterogeneity in attention deficit hyperactivity disorder (ADHD): gender, psychiatric comorbidity, and maternal ADHD. J Abnorm Psychol 104:334–345. Faraone SV, Biederman J, Keenan K, Tsuang MT (1991). A family-genetic study of girls with DSM-III attention deficit disorder. Am J Psychiatry 148:112–117. Frick PJ, Kamphaus RW, Lahey BB, et al. (1991). Academic underachievement and the disruptive behavior disorders. J Consult Clin Psychol 59:289–294. Gaub M, Carlson CL (1997). Gender differences in ADHD: a meta-analysis and critical review. J Am Acad Child Adolesc Psychiatry 36:1036–1045. Giedd JN, Castellanos FX, Rajapakse JC, Vaituzis AC, Rapoport JL (1997). Sexual dimorphism of the developing human brain. Prog Neuropsychopharmacol Biol Psychiatry 21:1185–1201. Guy W (1976). Subject’s Treatment Emergent Symptom Scale. In: Assessment Manual for Psychopharmacology. Guy W, ed. Washington, DC: US Government Printing Office, pp 347–350. Heptinstall E, Taylor E, Sonuga-Barke EJ, Sandberg S, Bowyer J (1998). Sex differences in the association of hyperactivity and conduct disorder. Presented at Royal College of Psychiatrists, Annual Scientific Meeting, London, January. Herjanic B, Campbell W (1977). Differentiating psychiatrically disturbed children on the basis of a structured interview . J Abnorm Child Psychol 5:127–134. Hollingshead AB (1975). Four Factor Index of Social Status. New Haven, CT: Yale University Department of Sociology. Hudziak JJ (1997). Identification of phenotypes for molecular genetic studies of common childhood psychopathology. In: Handbook of Psychiatric Genetics. Blum K, Noble EP, eds. Boca Raton, FL: CRC Press, pp 201–217. Mannuzza S, Gittelman R (1984). The adolescent outcome of hyperactive girls. Psychiatry Res 13:19–29. Pelham WE Jr, Walker JL, Sturges J, Hoza J (1989). Comparative effects of methylphenidate on ADD girls and ADD boys. J Am Acad Child Adolesc Psychiatry 28:773–776. Rapport MD, DuPaul GJ, Stoner G, Jones TJ (1986). Comparing classroom and clinic measures of attention deficit disorder: differential, idiosyncratic, and dose-response
ADHD in girls
249
effects of methylphenidate. J Consult Clin Psychol 54:334–341. Rhee SH, Waldman ID, Hay DA, Levy F (in press). Sex differences in genetic and environmental influences on DSM-III-R attention-deficit hyperactivity disorder (ADHD). J Abnorm Psychol. Rosvold HE, Mirsky AF, Sarason I, Bransome ED, Beck LH (1956). A continuous performance test of brain damage. J Consult Psychol 20:343–350. Safer DJ, Zito JM, Fine EM (1996). Increased methylphenidate usage for attention deficit disorder in the 1990s. Pediatrics 98:1084–1088. SAS Institute (1996). SAS® Companion for the Microsoft Windows Environment. Cary, NC: SAS Institute Inc. Shaffer D, Gould MS, Brasie J, et al. (1983). A children’s global assessment scale (CGAS). Arch Gen Psychiatry 40:1228–1231. Spencer T, Biederman J, Wilens T (1996). Responders and nonresponders (letter). J Am Acad Child Adolesc Psychiatry 35:1569–1570. Spitzer RL, Endicott J (1982). Schedule for Affective Disorders and Schizophrenia (SADS). New York: New York State Psychiatric Institute. Swanson JM, Sergeant JA, Taylor E, Sonuga-Barke EJS, Jensen PS, Cantwell DP (1998). Seminar: attention-deficit hyperactivity disorder and hyperkinetic disorder. Lancet 351:429–433. Szatmari P (1992). The epidemiology of attention-deficit hyperactivity disorder. Child Adolesc Psychiatr Clin North Am 1:361–371. Tannock R, Ickowicz A, Schachar R (1995). Differential effects of methylphenidate on working memory in ADHD children with and without comorbid anxiety. J Am Acad Child Adolesc Psychiatry 34:886–896. Taylor E, Heptinstall E, Sonuga-Barke EJ, Sandberg S (1998). Sex differences in the prevalence of hyperactivity. Presented at Royal College of Psychiatrists, Annual Scientific Meeting, London, January. Ward MF, Wender PH, Reimherr FW (1993). The WURS: a rating scale to aid in the retrospective diagnosis of attention deficit disorder in childhood. Am J Psychiatry 150:885–890. Wechsler D (1974). Manual for the Wechsler Intelligence Scale for Children-Revised. New York: Psychological Corporation. Wechsler D (1991). Wechsler Intelligence Scale for Children-Third Edition. San Antonio, TX: Psychological Corporation. Werry JS, Sprague RL, Cohen MN (1975). Conners’ Teacher Rating Scale for use in drug studies with children: an empirical study. J Abnorm Child Psychol 3:217–229. Woodcock RW, Johnson BB (1977). Woodcock-Johnson Psychoeducational Battery. Allen, TX: DLM Teaching Resources. Woodcock RW, Johnson MB (1989). Woodcock-Johnson Psycho-Educational BatteryRevised. Allen, TX: DLM Teaching Resources.
PART III: ATTENTION DEFICIT-HYPERACTIVITY DISORDERS
13 Stimulant Treatment for Children: A Community Perspective Adrian Angold, Alaattin Erkanli, Helen L.Egger, and E.Jane Costello
Objective: To examine the use of prescribed stimulants in relation to research diagnoses of attention deficit-hyper activity disorder (ADHD) in a community sample of children. Method: Data from 4 annual waves of interviews with 9- to 16-year-olds from the Great Smoky Mountains Study were analyzed. Results: Over a 4-year period, almost three quarters of children with an unequivocal diagnosis of ADHD received stimulant medications. However, girls and older children with ADHD were less likely to receive such treatment. Most children with impairing ADHD symptoms not meeting full criteria for DSM-III-R ADHD did not receive stimulant treatment. Stimulant treatment in this group was significantly related to the level of symptoms reported by parents and teachers and was much more common in individuals who met criteria for oppositional defiant disorder. The majority of individuals who received stimulants were never reported by their parents to have any impairing ADHD symptoms. They did have higher levels of nonimpairing parent-reported ADHD symptoms, higher levels of teacher-reported ADHD symptoms, and interviewer-observed ADHD behaviors, but these typically fell far below the threshold for a DSM-III-R diagnosis of ADHD. Conclusions: In this area of the Great Smoky Mountains, stimulant treatment was being used in ways substantially inconsistent with current diagnostic guidelines. (J Am Acad Child Adolesc Psychiatry, 2000, 39(8):975–984.) Key Words: stimulants, attention deficit-hyperactivity disorder, treatment, community, epidemiology.
INTRODUCTION Stimulants (particularly methylphenidate) are the first-line medications for attention deficit-hyperactivity disorder (ADHD). Numerous controlled trials over many years have established beyond a doubt the efficacy of these drugs in reducing hyperactivity and improving attentiveness, at least in the short term. Methylphenidate has also proved to be a very safe and generally well-tolerated drug, and clear guidelines are available for
Annual progress in child psychiatry and child development 2000–2001
252
determining optimal dosage levels and avoiding side effects (Goldman et al., 1998). Studies based on prescription and prescriber records have shown that during the first half of this decade consultations for overactivity and attentional problems approximately doubled, whereas the number of children receiving stimulant treatment increased about 2.5-fold (Safer et al., 1996). Estimates of annual national utilization rates for mid-1995 were 3% to 4% of 5- to 14-year-olds and 2.5% to 3% of 5- to 18year-olds. At least part of the increased consumption is accounted for by increasing numbers of physician visits for ADHD and the use of methylphenidate by individuals over longer periods of time and at older ages than was previously the case (Swanson et al., 1995). In addition, more girls are receiving stimulants (Safer & Krager, 1988; Safer et al., 1996). There are also dramatic differences in prescription rates among geographic areas, probably resulting from variability in the prescribing habits of physicians (Morrow et al., 1998; Rappley et al., 1995). A recent American Medical Association Council report (Goldman et al., 1998), citing evidence that rates of stimulant prescription were not higher than the general population prevalence of ADHD, concluded that “there is little evidence of widespread overdiagnosis or misdiagnosis of ADHD or of widespread overprescription of methylphenidate” (p. 1100). However, the report also noted that “few data exist on actual practice habits in terms of what diagnostic criteria (if any) are used by clinicians, how they are applied, or exactly what a minimally satisfactory level of investigation entails” (p. 1105). In other words, we know little about the degree to which the increased use of methylphenidate represents appropriate use of an effective medication for children with ADHD, and there are some indications that many children receiving stimulants may not meet current diagnostic criteria for ADHD (Jensen et al., 1989; Wolraich et al., 1990). From the perspective of “evidence-based medicine,” we should expect that stimulant medications will be prescribed to those with ADHD and withheld from those who do not suffer from this disorder. Only one study has addressed this question directly (Jensen et al., 1999): In a cross-sectional survey of four U.S. communities, only 12% of children meeting full DSM-III-R criteria for ADHD had received stimulants during the preceding year. These children accounted for half of the total of just 16 medication prescriptions identified. The other eight prescriptions were to children who had “quite high levels of ADHD symptoms.” In this report we use data collected between November 1992 and November 1996 to examine the relationship between clinical status and stimulant medication use in an 11county region in the Great Smoky Mountains.
METHOD Sample The Great Smoky Mountains Study is an ongoing, longitudinal study of the development of psychiatric disorders and need for mental health services in rural and urban youths. Full details of the study design can be found elsewhere (Costello et al., 1996). Briefly, a representative sample of 4,500 children and adolescents aged 9, 11, and 13
Stimulant treatment for children
253
years, recruited through the Student Information Management System of the public school systems of 11 counties in western North Carolina, was selected using a household equal probability design. A screening questionnaire, consisting mainly of questions about behavioral problems, was administered to a parent (usually the mother), by telephone or in person. All children scoring above a predetermined cutpoint, plus a 1-in-10 random sample of the rest, were recruited for detailed interviews. In addition, an oversample of all 9-, 11-, and 13-year-old American-Indian children and adolescents (n= 431) living in the area were recruited for the interview phase. Three hundred seventy-five American Indian children and adolescents took part in the study. The overall response rate was 80% (n=1,422). The lower end of the age range was set at 9 because 8 and 9 are the youngest ages at which child self-reports of psychiatric symptoms on the diagnostic interview used are reliable. The response rate at the first wave was 80% (n=1,422). Contact with families was maintained by annual interviews, and by telephone every 3 months. In all, the sample included 4,964 annual observations of the 1,422 subjects. Thus across four waves, 70% (80%×87%) of all interviews with eligible subjects were completed. Table 13.1 shows the age and sex distributions of these observations. At the end of 4 years, information was available on the age range 9 through 16, with overlapping data from two age cohorts at ages 11, 12, 13, and 14.
TABLE 13.1 Age and Sex Distributions of the Observations From the Great Smoky Mountains Study
Age Female Male Total a 195 239 434 9 (w1) 10 (w2) 210 258 468 11 (w1, w3) 369 493 862 12 (w2, w4) 395 485 880 13 (w1, w3) 353 452 805 14 (w2, w4) 369 439 808 15 (w3) 189 197 386 16 (w4) 152 169 321 Total 2,232 2,732 4,964 awN indicates the waves in which children of that age were interviewed. For example, (w1, w3) indicates that some children were 11 at their wave 1 interview, whereas others were 11 at their wave 3 interview. Measures Child and Adolescent Psychiatric Assessment. The Child and Adolescent Psychiatric Assessment (CAPA) (Angold and Costello, 1995; Angold et al., 1995) is a psychiatric interview for children aged 9 and older and their parents; it elicits information about symptoms that contribute to a wide range of diagnoses. The CAPA combines the characteristics of an “interviewer-based” and a “respondent-based” interview (Angold et al., 1995). Like respondent-based interviews, the CAPA uses a highly structured protocol,
Annual progress in child psychiatry and child development 2000–2001
254
with required questions and probes. However, as in an interviewer-based interview, the onus throughout is on the interviewer to ensure that subjects: (a) understand the question being asked; (b) provide clear information on behavior or feelings relevant to the symptom; and (c) have the symptom at a prespecified level of severity as defined in an extensive glossary. When symptoms are reported, their dates of onset are also collected, to determine whether they meet the symptom overlap and duration criteria for the various DSM diagnoses. A 3month “primary period” is used rather than a longer period, because shorter recall periods are associated with more accurate recall (Angold et al., 1996). Diagnoses and symptom scales are generated by computer algorithms. All diagnoses, except for ADHD, are based on information from both the parent and child. The diagnosis of ADHD is based on reports from the parent interview only, because of the poor validity of child-based ADHD ratings. Psychosocial impairment secondary to psychiatric symptoms in 17 areas of functioning related to life at home, at school, and elsewhere was also rated according to a series of definitions and rules specified in the CAPA glossary and the interview schedule. In general, some decrement in actual function had to be described for a positive rating to be given (see Angold et al., 1995, for a full description of the concept of impairment implemented in the CAPA). Briefly, having completed the symptom part of the interview, the interviewer reviewed with the subject the areas of positive symptoms, which are divided into 19 symptom groups (e.g., ADHD symptoms or depressive symptoms). Then for each of these areas the subject was questioned about whether those symptoms had resulted in impairment on a symptom group by symptom group and area of impairment by area of impairment basis. Note that there was no requirement that symptoms in any group should meet any sort of diagnostic criteria. A single symptom could be the basis for an impairment coding related to that symptom’s group. Thus, impairment resulting from diagnostically “sub-threshold” symptoms was often coded, as we see in the following. Once an impairment had been identified, interviewers were required to question the participants about what aspects of their symptoms had led to that impairment. Diagnostic 1-week test-retest reliabilities (κ values) for child self-reports were 0.55 for conduct disorder, 0.90 for major depression, 0.85 for dysthymia, 0.74 for overanxious disorder, 0.79 for generalized anxiety disorder, 1.0 for substance abuse/dependence, and 0.64 for post-traumatic stress disorder (Angold & Costello, 1995; Costello et al., 1998). A test-retest study of the impairment section of the child CAPA showed an intraclass correlation coefficient of 0.76 (Angold & Costello, 1995). The construct validity of the CAPA has been well supported (Angold & Costello, 2000). Parent Report-Based ADHD Diagnostic Groups. Because ADHD is a chronic disorder, with fluctuating levels of symptoms, and because we have no pathognomonic marker of its presence at any point in time, we regarded ADHD as being definitely present if at any of the 4 waves the child met full DSM-III-R criteria for ADHD by parent reports. DSMIII-R criteria were used because data collection began in 1992, before the advent of the DSM-IV. An additional group had parent reports of ADHD symptoms associated with functional impairment (at home, school, or elsewhere) but did not meet the full DSM-IIIR criteria for ADHD. We regarded these as having ADHD-not otherwise specified (NOS), as they had “prominent symptoms of inattention or hyperactivityimpulsivity that
Stimulant treatment for children
255
do not meet criteria for ADHD” (American Psychiatric Association, 1994). The third group consisted of all children who had neither ADHD nor ADHD-NOS (although they could have any other diagnoses). We used parent reports of ADHD symptoms to form these diagnostic groupings because we did not have full diagnostic information from any other source (e.g., the schools) and because parent reports form the core part of the assessment of ADHD in most “real world” clinical settings. Parent-Reported mean Four-Wave ADHD Symptom Count Fourteen symptoms count toward a DSM-III-R diagnosis of ADHD. For each child, we computed a mean count of ADHD symptoms across the four waves of observation, by summing the count of symptoms at each wave and dividing by the number of waves of observation for that child. This scale, therefore, had a range from 0 to 14 and could have scores of less than 1. For instance, a child whose parent reported a single ADHD symptom at just one of four interviews would receive a mean 4-wave symptom count of .25. Interviewer Observations of Overactivity, Distractibility, and Tics. The interviewers also made observational ratings of overactivity and distractibility during the interview with the child. A dichotomous rating was generated indicating whether or not (1/0) the child had ever been observed as being overac-tive or distractible at any of the interviews. These ratings provided an opportunity to determine whether a clinician might conclude that a child was hyperactive or distractible regardless of parent or teacher reports. Interviewers were trained to detect the presence of observable motor and phonic tics during the interview, and so codings of the observable presence of tics were available. Prescribed Stimulant Use. The CAPA also includes questions about use of medications for emotional or behavioral problems. If any medication use was reported by the child or parent, the interviewer recorded the name of that medication as reported and then asked to see the medication container to check what was actually being received. Stimulant medications were identified from a list of currently available stimulant medications and coded as such. No data on dosage were collected. The independent parent and child reports of stimulant use agreed closely (κ=.88). For the purposes of this article, stimulant users were those who reported taking prescribed stimulants at any of the four waves of assessment by parent or child report. Thus individuals had four “chances” to report stimulant use in any one of the four 3-month “windows” provided by the CAPA. Though this means that a 1-year period of observed time was covered, it is important to note that the resulting rates of stimulant use do not represent an approximation of the annual prevalence of stimulant use, because that “year of observation” was spread over 39 months of real time. Subjects receiving psychoactive medications were asked at each wave how long they had been taking these medications. We determined the longestduration report of the four possible reports and coded this as the duration of treatment. Teacher Reports from the Teacher’s Report Form. Information on school performance and problems was reported at each wave from wave 2 through 4 by three teachers per wave per child using the Teacher’s Report Form (TRF) (Achenbach, 1991a, 1991b). Ten items on the TRF can be regarded as corresponding reasonably directly to DSM-III-R or DSM-IV ADHD symptoms (item numbers: 4, 8, 10, 15, 17, 22, 41, 53, 78, 93). From each TRF we summed the item scores on these 10 items. We then took the average score from all the TRFs for each child across teachers and waves to produce a single average
Annual progress in child psychiatry and child development 2000–2001
256
TRF-based ADHD score for each child (range=0 to 20; mean=3.4). In summary, CAPAs were completed with the child and a parent at each annual wave, whereas TRFs were available only for waves 2 through 4. Data Analysis The principal statistical procedure used is logistic regression with the outcome variable being the use or nonuse of stimulants. For the most part, the three diagnostic groups are treated separately because: (a) we expected that there would be different predictors of treatment in each of them; and (b) the implications of the findings are different in each group. The presence of repeated measures and screen-stratified sampling required the use of weighted analyses to generate unbiased population parameter estimates and of “sandwich” type variance corrections (Diggle et al., 1994; Pickles et al., 1995) to produce appropriate confidence intervals and p values. These were obtained using generalized estimating equations implemented in SAS PROC GENMOD. We will first present data on the prevalence rates of ADHD and ADHDNOS and stimulant use and the duration of stimulant use. These analyses will be followed by a series of logistic regressions, including a variety of hypothesized predictors of receiving stimulant medications: (a) age at entry into the study (with younger children being more likely to receive stimulants); (b) gender (with boys being more likely to receive stimulants); (c) being below the federal poverty level at one or more assessments (with the poor being more likely to receive stimulants); (d) being an American Indian (with this minority group expected to be less likely to receive stimulants); (e) the average ADHD symptom count across the four observations by parent report; (f) the average teacherreported ADHD score (with the more severely symptomatic being more likely to receive stimulants); (g) psychosocial impairment due to non-ADHD symptoms at any point across the four waves (with those with impairing comorbid symptoms being more likely to receive stimulants); (h) interviewer observation that the child had been overactive, distractible, or inattentive at any of the 4 waves of interviews; and the presence of a (i) depressive, (j) anxiety, (k) oppositional defiant, or (1) conduct disorder diagnosis at any of the four waves (with those meeting criteria for any of these specific diagnoses being more likely to receive stimulants). An initial model was fit by first including all the possible predictor variables and iteratively removing items with the smallest nonsignificant (p>0.05) effects. When only significant effects (p<0.05) were left, we entered the previously excluded terms one at a time to determine whether they would be significant in the reduced model. When such a term was found, all the other excluded terms were again examined to determine whether they would have a significant effect in the new model. The final models consisted of all terms found to be significant following this procedure. Model selection, therefore, was not dependent on unreliable automated “stepwise” procedures. These procedures were all based on variables summarizing the child’s status across the four waves of observation, so each child was represented once (for a total sample n of 1,420 based on 4,964 individual observations). The final set of analyses addresses several questions that require analysis at the level of the individual observation.
Stimulant treatment for children
257
RESULTS Prevalence of ADHD and Stimulant Use Of the interviewed children, 3.4% (n=92) were regarded as having definite DSM-III-R ADHD. Boys were more likely to have definite ADHD (5.3% versus 1.5% in girls; odds ratio [OR]=3.7, 95% confidence interval [CI] 2.1 to 6.8, p<0.0001). These prevalence rates are consistent with those from other general population studies (Goldman et al., 1998). An additional 2.7% (N interviewed=63) had ADHD-NOS. Boys were also overrepresented in this group (4.4% versus 1.0% in girls; OR=4.6, 95% CI 2.2 to 9.5, p<0.0001). Thus we had 6.2% with one or the other ADHD diagnosis (9.7% of boys and 2.5% of girls; OR=4.2, 95% CI 2.6 to 6.9, p<0.0001). The individual prevalence rates ofADHD and ADHD-NOS do not sum exactly to the prevalence of combined ADHD or ADHD-NOS category because these are all weighted estimates. One hundred sixty-eight (7.3%) of the children had received stimulants at some point during the four waves of observation. Thus, more than twice as many children in the community were estimated to have received stimulants as received a full ADHD diagnosis, and even when ADHD-NOS was included, the number of children with parentreported ADHD was smaller than the number who received stimulants. To give an indication of the estimated relative sizes of these groups in the community, Table 13.2 provides the estimated numbers (weighted N’s) of children aged 9, 11, and 13 in 1993 (the year we selected the screened and interviewed samples) belonging to each group. The actual n’s in the interviewed groups are not helpful here, because we used a screen that oversampled those with ADHD, and so deliberately biased the proportions in each group. It can be seen that in the community, the majority of those who received stimulants never met criteria for ADHD or ADHD-NOS. Of those who received stimulants at any point during the study, 34% met full ADHD criteria, 9% had ADHDNOS, and 57% never had parent-reported impairing ADHD symptoms of any sort.
TABLE 13.2 Percentages of Girls and Boys Receiving Stimulants by ADHD Diagnostic Group
% Received Stimulants (Estimated n in 11 Counties) Female Male Total Neither ADHD nor ADHD-NOS 1.8 (97) 7.3 (385) 4.5 (482) ADHD-NOS 25.2 (14) 22.2 (57) 22.8 (73) ADHD 41.3 (34) 80.4 (249) 72.2 (283) Note: ADHD=attention deficit-hyperactivity disorder; NOS=not otherwise specified.
Annual progress in child psychiatry and child development 2000–2001
258
Stimulants and Diagnosis Table 13.2 shows the proportions of individuals who received stimulants by diagnostic group and sex. Although nearly three-quarters of those who met full ADHD criteria received stimulants at some point, more than three quarters of those with ADHD-NOS did not (OR=8.8, 95% CI 3.6 to 21.4, p<0.0001). However, those with ADHD-NOS were more likely to receive stimulants than those who had no ADHD diagnosis (OR=6.3, 95% CI 2.8 to 13.8, p<0.0001). Duration of Treatment The mean durations of treatment were 50.4 (SD=25.0) months for the ADHD group, 50.5 (SD=16.4) months for the ADHD-NOS group, and 40.3 (SD=29.9) months for the nonADHD/non-ADHD-NOS group. Thus, the average duration of treatment was greater than 3 years in all groups. Characteristics of Those Who Received Stimulants by Diagnostic Subgroup Table 13.3 shows the characteristics of stimulant users and nonusers in the 3 diagnostic groups in terms of the 12 potential predictors of stimulant use described in the following. Table 13.4 shows the results of logistic regressions performed on these data. Potential predictors that were without significant effects in any diagnostic group are not included in Table 13.4.
TABLE 13.3 Descriptive Statistics for the Relationships Among 12 Potential Predictors of Stimulant Use in Each of the Three Diagnostic Groups
Children With Neither Children With ADHDChildren With ADHD nor ADHD-NOS NOS (N=63 observed over (N=92 observed waves) four waves) (N=1,265 observed over four waves) Possible No Stimulanttreated No Stimulanttreated No Stimul (N Predictor stimulants (N=86) stimulants (N=24) stimulants (N=1,179) (N=39) (N=34) Male (%) 47.8 79.9 82.8 80.3 55.6 In Youngest 34.1 48.7 59.5 22.6 30.8 age cohort (%) In poverty 34.0 32.1 56.3 32.1 54.6 (%) American 3.6 4.8 3.0 10.2 5.5 Indian (%) Median 1.6 10.0 7.5 7.0 6.5
Stimulant treatment for children
259
teacher ADHD symptom score over three waves Median N 0 0.8 1.5 2.5 4.3 of parentreported ADHD symptoms over four waves Impairment 46.3 84.1 99.5 100 85.4 owing to non-ADHD symptoms over four waves (%) Interviewer10.1 49.8 44.1 52.1 42.7 observed ADHD symptoms (%) Diagnosis 4.2 8.1 4.4 27.9 43.2 of ODD (%) Diagnosis 5.5 13.9 22.4 20.0 25.8 of CD (%) Diagnosis 3.7 9.7 6.1 9.9 23.6 of depression (%) Diagnosis 9.0 5.5 9.6 9.9 29.9 of anxiety disorder (%) Note: ADHD=attention deficit-hyperactivity disorder; CD=conduct disorder; NOS otherwise specified; ODD=oppositional defiant disorder.
Annual progress in child psychiatry and child development 2000–2001
260
TABLE 13.4 Results of Logistic Regressions of Stimulant Use on 12 Potential Predictors by Diagnostic Group
Possible Predictor Gender
Children with Neither Children with ADHD norADHDADHD-NOS NOS (N=1,265 observed over four (N=63 observed over four waves) waves) — —
Children with ADHD (N=92 observed over four waves) 6.4 (2.1–19.1) *** — 0.23 (0.10–0.55) *** — — 4.5 (1.04–19.6)* —
Age at wave 1 (age — cohort) Poverty 0.38 (0.15–0.93)* American Indian 2.8 (1.2–6.7)* versus the rest Average teacher 1.2 (1.1–1.4)** — — ADHD symptom score over four waves Average N of 1.7 (1.1–2.2)* 2.6 (1.2–6.0)* — parentreported ADHD symptoms over four waves Impairment owing to 3.0 (1.1–8.3)* — non-ADHD symptoms over four waves Interviewer-observed 4.4 (1.6–11.6)** — ADHD symptoms — 11.6 (2.6–48.3)** — Diagnosis of ODD Note: Values represent odds ratios (and 95% confidence intervals). Potential predictors that were without significant effects in any diagnostic group are not included. ADHD=attention deficithyperactivity disorder; NOS=not otherwise specified; ODD=oppositional defiant disorder. *p<.05;** p<.01; *** p<.001.
Full ADHD. Of the factors possibly associated with stimulant use described in the preceding, in the final models only two had a significant effect on the probability of receiving stimulants in the ADHD group. Boys and younger children were more likely to receive stimulants. ADHD-NOS. In the ADHD-NOS group there were significant effects of level of ADHD symptoms by parent report, the presence of oppositional defiant disorder (ODD), and being an American Indian. Sixty-five percent of those who met criteria for ODD received stimulants, compared with 18% of the rest. Fifty percent of American Indians in this group received medication, compared with 21% of the rest.
Stimulant treatment for children
261
Neither ADHD norADHD-NOS. Figure 13.1 shows that among those who had neither ADHD nor ADHD-NOS, an average of less than one parent-reported symptom per wave of observation was sufficient to substantially elevate the probability of receiving stimulant medication and that higher levels of symptoms were associated with still higher probabilities of receiving medication. The median four-wave ADHD symptom count for the treated group was 2, compared with a median of 0 for untreated children (81.4% scored 0). More than a quarter (29.3%) of those without ADHD or ADHD-NOS receiving stimulants had an ADHD symptom score of 0. In other words, at no time over 4 waves of observation did their parents report that they had any ADHD symptoms. Of the whole population (including those who met criteria for ADHD or ADHD-NOS), 7.6% had a score of 2 or more and only half (47.8%) of these received stimulants.
Figure 13.1. Percentage who used stimulants in those without attention deficithyperactivity disorder (ADHD) or ADHD-NOS. Teacher reports also predicted stimulant use in children without a diagnosis of ADHD or ADHD-NOS. Stimulant users had a median score of 10.0 on this scale, compared with a median of 1.6 for the nonusers. Clearly the stimulantusing group showed ADHD-type problems at school, but it must also be noted that 5.7% of the non-stimulant users also scored at or above 10 on this scale. Those scoring above 10 constituted 8.1% of the whole
Annual progress in child psychiatry and child development 2000–2001
262
population, and only 30% of them received stimulants. Also, by definition, half of the stimulant users scored less than 10. Stimulant users were also substantially more likely to have been rated by the interviewers as being more overactive or distractible, and they were more likely to have impairment resulting from non-ADHD symptoms (by definition this group could not have impairing ADHD symptoms). In addition, those living below the federal poverty level were less likely to receive stimulants and American Indians were nearly three times as likely to receive stimulants. We now turn to a series of questions that require regression analysis using one or more of the interview observations separately, rather than depending on variables that summarize the child’s status across all four waves. Had the Non-ADHD/Non-ADHD-NOS Group Who Had Used Stimulants Been “Cured” by Them? One obvious possible explanation of the fact that most treated individuals did not meet criteria for ADHD was that stimulant treatment had been so successful that the symptoms had disappeared. A test of this proposition was offered by the fact that 56% of the treated non-ADHD individuals began medication after the first wave interview, thus providing the opportunity for a pre- or posttreatment comparison. By definition, these individuals had not met ADHD or ADHD-NOS criteria prior to stimulant treatment—if they had, they would have been allocated to one of those two groups. At the assessment prior to that in which they began stimulant treatment, 71% had no parent-reported ADHD symptoms, 27% had just one, and 2% had three. Their counts of ADHD symptoms the following year were as follows: 56% had no symptoms, 2% had one, 2% had two, 38% had three, and 2% had 11. Thus, low symptom counts in those started on stimulants during the course of the study cannot be accounted for by their curative effect. Treatment Side Effects Children who were currently being treated with stimulants had much higher rates of interviewer-observed tics than those who were not. When children were taking stimulants, 3.9% of them had tics, compared with 0.4% of observations of children never treated with stimulants (OR=3.2, 95% CI 1.7 to 6.0, p=0.0004) and 0.16% of observations of children who were receiving stimulant treatment at some point but not during the 3 months prior to the observation (OR=24.6, 95% CI 2.7 to 220.6, p=0.004). Secular Change in the Use of Stimulants As shown in Figure 13.2, the prevalence of stimulant use changed little from wave to wave. However, this is not very informative, because the children aged from wave to wave, and it is well known that the numbers of children meeting full criteria for ADHD falls in adolescence. Therefore, we corrected the effect of wave for changes in age and symptom levels (by parent report). The corrected rates by wave are seen in the solid line in Figure 13.2, which shows that there was a significant increase in the corrected
Stimulant treatment for children
263
probability of receiving stimulants over the course of the study (OR=1.3, p=0.01).
Figure 13.2. Secular trends in stimulant use. A Single-Wave Analysis for Comparison with Jensen et al. (1999) The results we have reported so far are dramatically different from those reported in the only other comparable study of stimulant use (Jensen et al., 1999). It is possible that this could be explained, in part, by our use of summary variables resulting from four waves of data collection, whereas Jensen et al. had only a single wave of observation. Therefore, we now consider just the observations made at wave 1, when the children were aged 9, 11, or 13, which represents the nearest approximation to Jensen and colleagues’ work that we could achieve. At wave 1, 1.8% met full criteria for ADHD (N=50), and, of these, 62% (N=28) had used stimulant medications during the preceding 3 months; 1.7% (N=38) had ADHD-NOS. Of these, 17.1% (n=9) had used stimulants during the preceding 3 months. Of children who had neither ADHD nor ADHD-NOS (N =1,241),
Annual progress in child psychiatry and child development 2000–2001
264
3.9% (N=80) had used stimulants during the preceding 3 months. Thus, 5.1% of children (N=117) had used stimulant medications during the preceding 3 months and only 22.1% of these met full criteria for DSM-III-R ADHD, 5.8% had ADHD-NOS, and the remaining 72.1% had neither ADHD nor ADHD-NOS. In other words, in a single wave as well as across four waves, we found the same general pattern of quite high levels of stimulant use among those with clear ADHD, but with most stimulant users not meeting even relaxed criteria for ADHD. A very similar pattern was found when we looked at waves 2 through 4 individually, so these results are not a peculiarity of wave 1.
DISCUSSION Among children identified by their parents as meeting full DSM-III-R criteria for ADHD, 72% received stimulants at some point during four annually assessed 3-month periods, as did 22% of those with ADHD-NOS, and 5% of those with neither ADHD nor ADHDNOS. The rate of stimulant treatment in this largely rural population was twice the rate of unequivocal parent-reported ADHD, and the majority of stimulant-treated children and adolescents did not meet criteria for either ADHD or ADHD-NOS. The propensity to treat children with stimulants increased across the four annual waves of observation of the study (covering the period between the end of 1992 and the end of 1996), thus paralleling the overall nationwide increase in the prescription of stimulants that has been documented through the mid-1990s. Although the prevalence of stimulant treatment in the non-ADHD/nonADHD-NOS group was relatively low, this group was much larger than the others, and so contributed the majority of treated individuals. Treated nonADHD/non-ADHD-NOS individuals did, however, have low levels of apparently nonimpairing ADHD symptoms, and half of them were also observed by interviewers to be overactive or inattentive. Their teachers rated them as having, on average, substantially more ADHD symptoms than nontreated individuals. However, many children not treated with stimulants had as many or more ADHD symptoms than those who had neither ADHD nor ADHD-NOS but were treated. Indeed, 13% of all children in the sample fell above the median parent- or teacherreported ADHD score of the non-ADHD/non-ADHDNOS treated group, but only 36% of these received stimulants. We doubt that anyone would agree that more than 13% of children have a disorder that merits stimulant treatment, and 35% of children who did receive stimulants did not even have this degree of ADHD symptoms. We should also note that current diagnostic guidelines indicate that a diagnosis of ADHD requires the presence of impairing ADHD symptoms in multiple settings, and there is little doubt that children in the non-ADHD/non-ADHD-NOS group were far from meeting these criteria. Other factors affecting the probability of treatment varied depending on the degree of ADHD symptoms observed. Thus, among those who met full criteria for ADHD, boys and younger children were much more likely to receive stimulants than girls and older children. This suggests undertreatment of the latter. The fact that level of ADHD symptoms, as indicated by either parent or teacher reports, or interviewer observation did not influence stimulant use in this group is not surprising, as they all had very high levels of ADHD symptoms.
Stimulant treatment for children
265
Among those with subthreshold impairing ADHD symptoms (ADHD-NOS), the number of ADHD symptoms was a significant predictor of stimulant use. However, the presence of a diagnosis of ODD had a similarly powerful effect. This might seem at first to indicate that ODD was being misdiagnosed as ADHD. However, the presence of ODD made no difference to the probability of stimulant treatment in those with neither ADHD nor ADHD-NOS, so it seems more likely that, among those with ADHD-NOS, the presence of ODD served to lower the threshold for stimulant treatment of ADHD symptoms, rather than being mistaken for them. Although family income had no significant effect on the probability of receiving stimulants in the ADHD and ADHD-NOS groups, those with neither of these diagnoses who were below the federal poverty level were less than half as likely to have received stimulants. We might interpret this as suggesting that poverty is a protection against the receipt of poorly indicated medications. There was no difference in the rate of use of stimulant medications between American Indians with ADHD and others with ADHD, but among those with ADHD-NOS and those with neither ADHD nor ADHD-NOS, American Indians were substantially more likely to receive stimulants. In the presence of ADHD-NOS this could be seen as a sign that the American-Indian population was better served in respect of stimulants, insofar as it can be argued that ADHDNOS is an indication for stimulant medication. However, it is worrisome that stimulants should more often be prescribed to American Indians in the apparent absence of such indications. It should be noted, however, that the majority of the American Indian population here was concentrated in a relatively small area, so we cannot tell whether the high rate of stimulant prescription resulted from an effect of ethnicity per se, or whether it reflects the particular prescribing habits of the physicians serving that area. Zito et al. (1998), using 1991 Maryland Medicaid administrative data, found that whites aged 5 to 14 years were 2.5 times more likely to be prescribed stimulant medications than African Americans, but there seem to be no comparable published data on American Indians. These results are dramatically different from those recently reported in the only other community study to have looked at rates of stimulant use in relation to diagnosis (Jensen et al., 1999). There, undertreatment of ADHD was the predominant problem. However, the prevalence of stimulant use during the preceding year in that study was only 1.2%— that is less than half of the estimated national utilization rates for 1995 (Safer et al., 1996). This is not unexpected given that the data for the study were collected during the first half of 1992 and the present study demonstrates an increasing propensity to prescribe between 1992 and 1996. So it may be that if Jensen et al. were to repeat their study, they would find a different picture. We also know that there is considerable variation in prescribing practices among physicians, and that could also be responsible for the differences between Jensen and colleagues’ findings and ours. Either way, both studies document a very unsatisfactory state of affairs in the relationship between ADHD and its treatment with stimulants in the community. Clinical Implications There is no doubt that stimulants are an efficacious treatment for ADHD, but little
Annual progress in child psychiatry and child development 2000–2001
266
evidence that they should be prescribed for other child and adolescent disorders. An evidence-based approach to the prescription of stimulants would, therefore, imply that those with ADHD should receive a trial of stimulants (and often remain on stimulants for long periods of time), whereas those who do not have ADHD should not be treated with long courses of stimulants. Our data suggest that this evidence-based expectation is not being fulfilled in the communities described here. Our findings are in line with analyses of treated population databases, showing that increasing numbers of children are now receiving stimulant medications, but go beyond that to suggest that current treatment practice in the community is far from optimal. There is no doubt that stimulants have positive effects on hyperactivity and attention problems associated with ADHD in double-blind treatment trials (Goldman et al., 1998), but such trials are normally conducted in major research centers using individuals who meet rigorous criteria for ADHD. However, individuals in our treated non-ADHD/non-AHDH-NOS group had a mean ADHD symptom count across four waves of observation of only one parentreported symptom. Such children are very different from those who have been included in stimulant efficacy trials, and it cannot be assumed that stimulants are efficacious in the treatment of such low levels of ADHD symptoms. Indeed, we found no reduction in parent-reported ADHD symptoms in prestimulant versus on-stimulant comparisons within individuals, though there was a significant increase in the rate of tics (a wellknown side effect of stimulants) with concurrent treatment. That most individuals received stimulants for a long time (a mean duration of more than 3 years) is also cause for concern. Together, these results present a troubling picture of a serious mismatch between need for stimulant treatment and the provision of such treatment. Limitations We should also note some limitations of the study. First, it was conducted in a single rural area, and in the light of evidence that there are widespread differences in treatment practices in different areas, no claim can be made that these results are representative of stimulant use across America. However, when widespread differences in practice are to be anticipated, “average” practice may not be a very useful index of real behavior. Given the problematic picture seen in this study, it seems that further investigation of this topic in other areas is merited. As we have seen, such studies need to address two questions: (a) the degree to which those with a clear indication for stimulant treatment receive it; and (b) the degree to which those who do not manifest such an indication receive stimulant medications. A second substantial limitation of this work is that we had no direct measures of the dosages of stimulants, or why these children had been started and maintained on such treatment, at the level of either patient or physician reports. Nor did we have previous treatment histories or detailed measures of what other nonpharmacological treatments the children might have been receiving. Future studies would benefit from the inclusion of measures of these aspects of practice, as well as the ability to generate true annual prevalence rates of stimulant use.
Stimulant treatment for children
267
ACKNOWLEDGMENTS From the Center for Developmental Epidemiology, Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, NC. This project was supported by grants from the NIMH (MH-48085) and NIDA (DA-11301) and by Center funding from the NIMH (MH-57761).
REFERENCES AchenbachTM (1991a). Child Behavior Checklist: Teacher’s Report Form. Burlington: University of Vermont Center for Children, Youth, & Families. Achenbach TM (1991b). Manual for the Teacher’s Report Form and 1991 Profile. Burlington: University of Vermont Department of Psychiatry. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Angold A, Costello EJ (1995). A test-retest reliability study of child-reported psychiatric symptoms and diagnoses using the Child and Adolescent Psychiatric Assessment (CAPA-C). Psychol Med 25:755–762. Angold A, Costello EJ (2000). The Child and Adolescent Psychiatric Assessment (CAPA). J Am Acad Child Adolesc Psychiatry 39:39–48. Angold A, Erkanli A, Costello EJ, Rutter M (1996). Precision, reliability and accuracy in the dating of symptom onsets in child and adolescent psychopathology. J Child Psychol Psychiatry 37:657–664. Angold A, Prendergast M, Cox A, Harrington R, Simonoff E, Rutter M (1995). The Child and Adolescent Psychiatric Assessment (CAPA), Psychol Med 25:739–753. Costello EJ, Angold A, Burns BJ, et al. (1996). The Great Smoky Mountains Study of Youth: goals, designs, methods, and the prevalence of DSM-III-R disorders . Arch Gen Psychiatry 53:1129–1136. Costello EJ, Angold A, March J, Fairbank J (1998). Life events and posttraumatic stress: the development of a new measure for children and adolescents. Psychol Med 28:1275–1288. Diggle PJ, Liang KY, Zeger SL (1994). Longitudinal Studies. Oxford, England: Claredon Press. Goldman LS, Genel M, Bezman RJ, Slanetz PJ (1998). Diagnosis and treatment of attention-deficit/hyperactivity disorder in children and adolescents. JAMA 279:1100– 1107. Jensen PS, Kettle L, Roper MS, et al. (1999). Are stimulants overprescribed? Treatment of ADHD in four US communities. J Am Acad Child Adolesc Psychiatry 38:797–804. Jensen PS, Xenakis SN, Shervette RE, Bain MW, Davis H (1989). Diagnosis and treatment of attention deficit disorder in two general hospital clinics. Hosp Community Psychiatry 40:708–712. Morrow RC, Morrow AL, Haislip G (1998). Methylphenidate in the United States, 1990
Annual progress in child psychiatry and child development 2000–2001
268
through 1995. Am J. Public Health 88:1121. Pickles A, Dunn G, Vazquez-Barquero J (1995). Screening for stratification in twophase (“two-stage”) epidemiological surveys. Stat Methods Med Res 4:73–89. Rappley MD, Gardiner JC, Jetton JR, Houang RT (1995). The use of methylphenidate in Michigan. Arch Pediatr Adolesc Med 149:675–679. Safer DJ, Krager JM (1988). A survey of medication treatment for hyperactive/inattentive students. JAMA 260:2256–2258. Safer DJ, Zito JM, Fine EM (1996). Increased methylphenidate usage for attention deficit disorder in the 1990s. Pediatrics 98:1084–1088. Swanson JM, Lerner M, Williams L (1995). More frequent diagnosis of attention deficithyperactivity disorder. N Engl J Med 333:944. Wolraich ML, Lindgren S, Stromquist A, Milich R, Davis C, Warson D (1990). Stimulant medication use by primary care physicians in the treatment of attention deficit hyperactivity disorder. Pediatrics 86:95–101. Zito JM, Safer DJ, dosReis S, Riddle MA (1998). Racial disparity in psychotropic medications prescribed for youths with Medicaid insurance in Maryland. J Am Acad Child Adolesc Psychiatry 37:179–184.
Part IV OTHER CLINICAL ISSUES
PART IV: OTHER CLINICAL ISSUES
14 The Altering of Reported Experiences Daniel Offer, Marjorie Kaiz, Kenneth I.Howard, and Emily S.Bennett
Objectives: The unreliability of human memory is well documented in the literature, yet psychiatrists and other mental health care professionals rely on patient self-report in history taking. This study provides new evidence from a longitudinal study of autobiographical memory and discusses implications for the development and implementation of appropriate treatment plans and goals. Method: Seventy-three mentally healthy 14-year-old males were studied in 1962. Sixty-seven of these subjects were reinterviewed face-to-face at age 48. Questions concerning areas of family relationships, home environment, dating, sexuality, religion, parental discipline, and general activities were asked in both interviews. Results: Significant differences were found between adult memories of adolescence and what was actually reported during adolescence. Accurate memory was generally no better than expected by chance. Conclusions: If the accurate memory of one’s past is not better than chance in the mentally healthy individual, even more care probably should be taken in obtaining accurate historical information in the medically, psychologically, or otherwise health-compromised individual. It would be more constructive to treat recollections as existential reconstructions. (J Am Acad Child Adolesc Psychiatry, 2000, 39 (6):735–742.) Key Words: autobiographical memory, adolescent experiences, longitudinal study.
INTRODUCTION Psychiatrists routinely rely on patient self-report to obtain a medical and psychological history, even though researchers have long known that memory is often an unreliable and inaccurate source of information (Lindsay & Read, 1994; McGovern et al., 1998; Thompson et al., 1996). Moreover, patients may remain highly confident about their recollections, even in the face of evidence to the contrary (Cohler, 1994). This study presents new evidence about significant differences between memories of generally mentally healthy males at adolescence and at middle adulthood. Lay people often think of memory as a sort of tape library or file drawer of their
The altering of reported experiences
273
personal histories and that the act of remembering is simply one of replaying a tape of the past. One implication of this metaphor is that memory is perfect and complete, requiring only the proper cues to allow people to retrieve accurate records of past experiences (Cohler, 1994). Many proponents of memory recovery therapies (particularly with patients who claim childhood sexual abuse trauma) emphasize the need to search for the appropriate triggers or retrieval cues that will enable clients to recover accurate memories that otherwise seem completely out of their reach. Blume (1990), Courtois (1991), and Maltz (1990) suggested such triggers include childhood pictures and memorabilia, guided imagery, hypnosis, age regression, dream work, and group work. Although Olio (1989) expressly discussed the reconstructive nature of remembering, the underlying assumption made by most of these investigators appears to be that, given the proper triggers, accurate records of the past will be located and reproduced. Actually, the conceptualization of human memory as being infallible is relatively new (Bartlett, 1983; Roediger, 1980, 1990). Memory research and clinical practice has made it clear that memories are far from perfect records of past events. What one remembers is based on inference-like reconstructive processes (Bonnano, 1990; Freud, 1899; Loftus & Loftus, 1980; Lynn & Payne, 1997; Rubin, 1996; Rubin & Schulkind, 1997; Spence, 1984, 1994) that result from a dynamic process shaped by the complex interaction of many variables. Such variables include length of time from the event (e.g., Linton, 1978; Wagenaar, 1986); frequency of the event (Brewer, 1988; Linton, 1982; Smith, 1952; Wagenaar, 1986; White, 1982); level of emotionality (deviation from neutrality in either pleasant or unpleasant directions) (Brewer, 1988; Christianson & Safer, 1996; Linton, 1982; Smith, 1952; White, 1982); one’s present expectancies, needs, and beliefs (e.g., Beck, 1967, 1976; Lynn & Payne, 1997; McGovern et al., 1998); and personal interpretation and value placed on events that affect how memories are encoded in the first place (Beck, 1976). In addition, there is a growing body of knowledge on the neural basis of autobiographical memory in general (e.g., Aggleton, 1991; Buckner, 1996; Johnson et al., 1993; Kapur, 1993; Mishkin et al., 1984; Nyberg et al., 1996; Petri & Mishkin, 1994; Squire, 1987, 1992; Tulving, 1989). For example, the neuropsychobiological study of specific types of amnesias can help to localize various memory functions. Some frontal lobe patients show an intact ability to create autobiographical “memories,” but they have severe problems finding and integrating facts and monitoring their recall against reality (Baddeley & Wilson, 1986). Over the course of a year, approximately 40% of individuals seeking medical services also report feelings of depression and unhappiness that have been shown to result in loss of ability to concentrate and poor memory (e.g., Baddeley & Wilson, 1986; Neisser & Fivush, 1994). Beck (1967) and Beck et al. (1979) clearly demonstrated that depressed patients exhibit a negative view of themselves, their past experiences, and the future. They further showed that depressed individuals demonstrate cognitive distortions through engaging in faulty information processing. Similarly, Wright and Morley (1995) found significant differences in types of memories accurately retrieved between patients with chronic pain versus control subjects with no pain. Pain patients retrieved memories most attributable to events in which they experienced chronic pain, and thus memories were distorted through their subjective experiences of pain.
Annual progress in child psychiatry and child development 2000–2001
274
The accuracy of an individual’s memory also has become the focus of psychology and psychiatry—and in society more generally—as a consequence of discussions and debates concerning recovered memories of childhood abuse (e.g., Goldstein & Farmer, 1992; Harvey & Herman, 1994; Loftus, 1993; Neisser, 1982; Neisser & Fivush, 1994; Pezdek, 1994: Pezdek & Roe, 1994; Schacter, 1994; Spence, 1984; Woodruff & Birren, 1972; Yarrow et al., 1970). During the past several years, there has been an explosion of cases in which adults undergoing psychotherapy claim to have recovered long-repressed memories of sexual abuse at the hands of parents or other family members. Those who believe that they have been wrongfully accused on the basis of false memories, of course, deny these accusations. It is often difficult to determine whether and in what sense individual memories are true or false. The extreme emotional stress of both the patients and their families highlights the importance memory plays in our lives, regardless of accuracy (e.g., Goldstein & Farmer, 1992; Harvey & Herman, 1994; Loftus, 1993; Neisser, 1982; Neisser & Fivush, 1994; Yapko, 1994). Woodruff and Birren (1972) found longitudinal change in perceptions of personality. Adolescents were tested in 1944 using the California Personality Inventory, which is designed to cover major themes of externality, normative regulation of impulse, intellectual ability, and socialization. They were retested 25 years later and there were no significant differences in personality scores, but the difference in subjective perceptions of how subjects thought they had answered the inventory in 1944 was statistically significant. Similarly, in a study of childhood recollection of both parents and their children, Yarrow et al. (1970) found that mothers tended to modify their recall of earlier characteristics of adult sons and daughters to conform to sex-linked stereotypes. Recollections of the child’s past personality characteristics were clearly transformed to comply with present perceptions of their adult children’s personalities. In sum, memory experts agree on many issues, including the fact that memory can be fallible, incomplete, malleable, and susceptible to external factors. Each of these points of agreement and the psychological principles that underscore them can be documented in a variety of ways (e.g., Bekerian & Goodrich, 1995; Kihlstrom, 1994; Neisser & Fivush, 1994). By its very nature, memory is a constructive process that is influenced by a wide range of cognitive and social events (e.g., Bartlett, 1932; Cialdini, 1993), including information that is provided during the encoding of the original event, during storage of that memory, and during the retrieval of that memory (e.g., Garry et al., 1994; Johnson & Raye, 1981; Johnson et al., 1993). Moreover, the use of techniques such as hypnosis to enhance retrieval can lead to major changes in an individual’s reported recall and in the confidence in the accuracy of their memories (e.g., Krass et al., 1988; McConkey & Sheehan, 1995). Individuals have been found to believe strongly in the accuracy of their hypnotically induced memories, even when these memories are inaccurate (Barnier & McConkey, 1992). These examples illustrate the enormous gap between the view of memory held by the general public and those who do research on memory. Memory is far from perfect and under certain circumstances can be surprisingly inaccurate (e.g., McGovern et al., 1998). The present investigation represents one part of a longitudinal study concerning the development from adolescence to middle adulthood of a group of essentially normal males. One of the goals was to investigate the stability of memory concerning perceptions
The altering of reported experiences
275
of events and relationships that had occurred during adolescence. This study determined how well the individual at age 48 accurately recalled subjective perceptions from his adolescence. This was accomplished by examining the differences between memories adults had of their teenage years and what they actually said when they were interviewed as adolescents.
METHOD Subjects This study is an extension into adulthood of a longitudinal study begun in 1962 to develop a psychological portrait of the normal adolescent male’s progress through the teenage years into young adulthood. The research findings from the high school years are described in The Psychological World of the Teenager (Offer, 1969), and the 4 post— high school years are described in From Teenage to Young Manhood: A Psychological Study (Offer & Offer, 1975). Subjects were 73 adolescent males selected from two Midwestern high schools during the first month of their freshman year in 1962. Primarily from middle- to upper-middleincome families, subjects were selected whose selfimage was within 1 SD of the mean on at least nine of the 10 Offer Self-image Questionnaire (OSIQ; Offer et al., 1992) scales. The OSIQ assesses the teenager’s self-image, personality adjustment, psychological status, socialization, sexuality, familial relationships, and coping. The boys had had at least a C average during eighth grade, they were not under psychiatric care, and their parents perceived that their sons were mentally healthy. Subjects and their par-ents were studied throughout high school and the following 4 post-high school years. School records were available to the researchers, and the home-room teacher completed two sets of ratings on the boys, one during the freshman year and one during the junior year. In 1991, a follow-up study was begun. As of 1997, 71 of the original 73 subjects were still living; two had died of cancer. Sixty-seven of these 71 subjects participated in the follow-up study. All subjects signed consent forms that allowed for publication of results and continued follow-up contact. The average age at the time of the interview was 48 years. In each case, an interviewer, blind to the responses that the subject had given in adolescence, conducted a structured, face-to-face, 4-hour interview. Extensive information about all psychological aspects of the subject’s life and events (i.e., work, family, relationships, income, education, sexuality, health, religion, and parenting) was obtained. Twenty-eight questions asked in adulthood were identical with questions asked in adolescence. The subjects were asked specifically to remember what they had said as adolescents, not how they would respond now. Each question was prefaced with: “When you were in high school….”
RESULTS Table 14.1 displays the percent endorsements for each item at the two time points: during
Annual progress in child psychiatry and child development 2000–2001
276
the high school years and during adulthood. What difference would it make if we used reports during the teenage years or reports about the teenage years given in adulthood? For example, 28% of the teenage boys said that they did not like high school and homework. During adulthood, however, 58% said that they remembered not liking school and homework. Similarly, in high school 24% said that they most enjoyed relationships with their peers. In adulthood, 52% thought that they had most enjoyed relationships with peers. Eighty-two percent of the boys reported that they were disciplined with physical punishment, whereas only 33% of the men stated that they had received such punishment. Obviously, there were significant discrepancies between these estimates of endorsement. Concordance was assessed for each question by counting the number of times a subject gave the same response at both ages and dividing by the number of subjects, that is, the observed concordance for an item was the proportion of subjects giving the same answer to that item at both time points. The expected concordance for an item was the product of the proportion of subjects who would have been expected by chance to give the same response at both time points. (Thus, if 60% of the subjects had answered yes at each time point, the expected concordance would be [0.6×0.6]+[0.4×0.4], or 0.50.) The statistical significance of the observed concordance was evaluated as follows: Pearson contingency coefficients were used to quantify the degree of dependence between responses to the 28 questions concerning events and relationships given at age 14 and responses to the identical questions at age 48. The Fisher exact test was then used to compute the probability of obtaining any particular association contrary to the null hypothesis (of a chance relationship) with p<0.05. Only three of the 28 contrasts attained this level of statistical significance.
TABLE 14.1 Percent Endorsement of Items at Ages 14 and 48 Years
Percent Endorsement Age 14 Age 48 Family relationships 1. What is (was) your mother’s best trait? Competence Relationship with subject Intelligence and knowledge Discipline Emotional responsiveness 2. What is (was) your father’s best trait? Competence Relationship with subject Intelligence and knowledge Discipline Emotional responsiveness 3. What is (was) your mother’s worst trait? Impulsivity
27 17 3 5 48
28 39 3 0 30
23 21 7 1 48
33 25 11 1 30
32
17
The altering of reported experiences
277
Nonaffectional (does not show feelings) Incompetence and unintelligence Relations with subject Discipline Habits 4. What is (was) your father’s worst trait? Impulsivity Nonaffectional (does not show feelings) Incompetence and unintelligence Relations with subject Discipline Habits 5. Who is (was) your mother’s favorite? I was Another sibling 6. Who is (was) your father’s favorite? I was Another sibling 7. Which parent do (did) you take after? Father Mother 8. Do (Did) you expect to earn more money than your father? Yes No 9. What is (was) the nicest thing about your home life? Physical comfort Emotional comfort Closeness with parents Other responses 10.What is (was) the worst thing about your home life? Physical comfort Emotional comfort Overt family conflict Sibling rivalry Decrease in self-esteem 11.What would you like (have liked) to change in your home life? Physical setting Emotional environment Relationship with parents Relationship with siblings Dependence-independence Nothing 12.Do (Did) your parents work on projects together?
24 2 14 9 19
2 11 23 19 28
41 27 4 7 7 14
26 11 2 21 12 28
14 86
30 70
28 72
26 74
71 29
49 51
70 30
65 35
9 61 21 9
19 55 23 3
40 11 13 16 20
15 50 33 2 0
27 12 19 13 7 22
14 33 24 6 6 17
Annual progress in child psychiatry and child development 2000–2001
Tes No 13.Which one of your parents makes (made) most of the decisions? Father Mother Dating and sexuality 14.Do (Did) girls like you? Yes No 15.Is (Was) it important for you to have a girlfriend? Yes No 16.Is (Was) it easy for you to get a date? Yes No 17.When is (was) sexual intercourse OK? Anytime during high school After high school Only for married people Religion 18.Is (Was) religion helpful to you? Yes No Discipline 19.Describe the discipline you receive(d) from your mother. Strict Not strict 20.Describe the discipline you receive(d) from your father. Strict Not strict 21.Is (Was) the discipline you receive(d) consistent? Yes No 22.Is (Was) the discipline you receive(d) upsetting to you? Yes No 23.Is (Was) the discipline you receive(d) unfair? Yes No 24.Do (Did) you receive physical punishment as discipline? Yes No 25.Is (Was) love withheld as punishment?
278
66 34
36 64
76 24
55 45
98 2
86 14
62 38
61 39
77 23
65 35
15 32
53
44 23 33
70 30
26 74
36 64
30 70
21 79
48 52
49 51
17 83
12 88
49 51
12 88
22 78
82 18
33 67
The altering of reported experiences
Yes No Activities 26.What do (did) you enjoy most? Physical outlets Relationships with peers Mental activity Family relationships 27.What do (did) you enjoy least? Physical work School and homework Peer relationships Other 28.Do (Did) your parents encourage you to be active in sports? Yes No
279
8 92
14 86
61 24 5 10
23 52 23 2
19 28 18 35
6 58 24 12
60 40
38 62
A review of Table 14.2 shows that the subjects’ recollections were about the same as would have been expected by chance in all but three items of the six content areas that were studied at both points in time: family relationships, home environment, dating and sexuality, religion, discipline, and activities. Responses and their accuracy did not vary according to the subject’s current mental health status at age 48. Whether the subject exhibited psychopathology did not affect the accuracy of recalled memories. The accuracy of recalled memories was uniformly poor. A closer look shows that the findings in seven out of eight items in the area of family relationships are about the same as would have been expected by chance. The single item that is significantly more accurate dealt with the subject’s future ability to earn money in comparison with that of his father. Subjects’ memories of home environment were no better than chance for all five questions, whereas in the area of dating and sexuality, only one item out of four memories was better than what would have been expected by chance. That item dealt with the importance of having a girlfriend in adolescence. There was one question about religion. Subjects’ memories were no better than chance when remembering whether they had stated as adolescents that religion was helpful. Regarding discipline, subjects did not remember any of the seven items accurately. Three items dealt with activities. Subjects remembered only one significantly more accurately than would have been expected by chance. That question recalled parental encouragement to be active in sports. Table 14.2 shows dramatically that with only three exceptions the observed and expected endorsements were almost identical, again illustrating that, in general, the accuracy of recalled memory is no better than chance.
Annual progress in child psychiatry and child development 2000–2001
280
TABLE 14.2 Percent of Responses Answered Items the Same at Both Ages 14 and 48 Years
Percent Consistent Endorsement Observed Expected by Chance Family relationships 1. What is (was) your mother’s best trait? 2. What is (was) your father’s best trait? 3. What is (was) your mother’s worst trait? 4. What is (was) your father’s worst trait? 5. Who is (was) your mother’s favorite? 6. Who is (was) your father’s favorite? 7. Which parent do (did) you take after? 8. Do (Did) you expect to earn more money than your father?* Home environment 9. What is (was) the nicest thing about your home life? 10. What is (was) the worst thing about your home life? 11. What would you (have) like(d) to change in your home life? 12. Do (Did) your parents work on projects together? 13. Which one of your parents makes/made most of the decisions? Dating and sexuality 14. Do (Did) girls like you? 15. Is (Was) it important for you to have a girlfriend?* 16. Is (Was) it easy for you to get a date? 17. When is sexual intercourse OK? Religion 18. Is (Was) religion helpful to you? Discipline 19. Describe the discipline you receive(d) from your mother. 20. Describe the discipline you receive(d) from your father. 21. Is (Was) the discipline you receive(d) consistent? 22. Is (Was) the discipline you receive(d) upsetting?
21 30 18 26 64 65 34 75
26 26 12 17 64 60 30 56
44
40
16
15
28
20
55
46
43
36
89a 64
85 54
79 34
61 31
45
40
25
56
20
51
56
50
48
51
The altering of reported experiences
281
23. Is (Was) the discipline you receive(d) 72 72 unfair? 24. Do (Did) you receive physical punishment 41 39 as discipline? 25. Is (Was) love withheld as punishment? 84 80 Activities 35 28 26. What [activities] do (did) you enjoy most? 27. What [activities] do (did) you enjoy least? 24 30 28. Do (Did) your parents encourage you to be 68 48 active in sports?* aEighty-nine percent is a very high level of agreement. Given the endorsement rates, however, 85% agreement would have occurred by chance. As shown in Table 1, 98% of the boys answered that girls like them; 86% of the men answered that girls liked them when they were teenagers. Thus, by chance we would expect (0.98′0.86)+(0.02′0.14)=85% concordance. The observed concordance of 89% was not statistically greater than the expected 85%. *p<0.05.
DISCUSSION Before discussing the results of this study, we want to remind the reader that our sample was originally drawn from middle- and upper-middle-class suburban high schools in the Midwest. We can think of no particular reason why the results would have been different for another sample, but they are most conservatively applicable to the population that was sampled here. This review of the 28 items clearly showed that the subjects did not recall their adolescence with any accuracy. The results would also seem to contradict previous studies concerning the importance of affect associated with accurate recall of autobiographical information. Level of emotionality has been found to be an important variable in accuracy of retrieved memories. Events initially rated as positive or negative were better remembered than events that were neutral in affect (e.g., Brewer, 1988; Linton, 1982; Smith, 1952; White, 1982). In this study, items that one would expect to have emotional significance such as type of discipline and relationships were not remembered any more accurately (except for importance of a girlfriend) than items without emotional overtones. The data presented corroborate how memories recalling events of the patient’s past may not be accurate regardless of the patient’s mental health status. Establishing the truth of autobiographical memory as well as recent memories of mental health care-related issues and treatments requires evidence from other corroborating sources (e.g., family members and medical records) to establish the validity of a patient’s memories. Of course, the memories of adults may be based on reevaluations of their adolescent experiences, that is, they may be accurate in a subjective sense. However, they do not represent accurate recall of how these experiences were evaluated during their teenage
Annual progress in child psychiatry and child development 2000–2001
282
years. Thus, our findings reinforce and place greater emphasis on the practice of using collateral sources to enhance the validity of relevant historical information. Even though a patient’s construction of the past is not necessarily related to the past as it actually happened, these constructions are important for the mental health care professional’s understanding of the current mental status of the individual. Many developmental studies have a longitudinal design. They investigate how a person’s personality unfolds and which kind of variables influence the person, when, and how much. Our study focused more on memories of the emotionally laden experience of adolescence as seen through the lens of 48year-olds. It is often said that adolescence is the period in the life cycle that is most difficult to see clearly. Our study demonstrated that this may indeed be so. However, a man’s reconstructions of his past, specifically of his adolescence, provides important information. It is the way the individual views his life. Understanding this view provides important insight about the individual, his feelings, and the nature of his relationships in the here and now. Whether it relates to other times and other places is another issue. In our case it did not. Clinical Implications The implications of these findings for psychiatrists and other mental health care providers who must routinely obtain historical and biographical information from the patient to develop and implement appropriate treatment plans and goals are significant. If accurate memory of past events and relationships is no better than chance for normal, mentally healthy individuals, we might expect that the reports of past experiences by people who are currently under stress (medically ill, psychologically disturbed, or otherwise compromised) would be even less accurate. Even so, it is not very likely that the information on psychological history given by a patient is accurate. The taking of a medical history needs to be conducted in the context of a trusting, caring relationship between the health care professional and the patient to find out what and how the patient thinks (e.g., Linton, 1978; Lynn & Payne, 1997; Spence, 1984). Patients who perceive their experience with health care professionals as warm and empathic are more likely to comply with treatment recommendations, invest themselves in the information-gathering aspect of medical care, and rate their overall experience with health care providers as positive (e.g., Daley & Zuckoff, 1998; Mitchell, 1998). Psychiatrists need to remember that inaccurate memories can be costly and dangerous and can lead to faulty conclusions and inaccurate diagnoses. For treatment adherence, this may mean that the true cause of a treatment failure may not be identified. Successful treatments may be judged ineffective; unnecessary procedures may be ordered to reevaluate a problem; and dose-response relationships may be miscalculated. Limitations Our sample was selected because the subjects scored in the average range (not too high and not too low) on entering high school. The sample was mostly white, suburban, and middle-class, and all subjects were male. A more diverse sample would help to determine
The altering of reported experiences
283
whether conclusions based on our sample may be generalizable.
ACKNOWLEDGMENT This research was made possible, in part, by a 3-year project grant from Thomas F.Pick, Chicago.
REFERENCES Aggleton JP (1991). Anatomy of memory. In: Memory Disorders: Research and Clinical Practice, Yanagihara T, Peterson RC, eds. New York: Marcel Dekker. Baddeley AD, Wilson B (1986). Amnesia, autobiographical memory, and confabulation. In: Autobiographical Memory, Rubin DC, Barclay CR, Smith TS, eds. Cambridge, England: Cambridge University Press. Barnier AJ, McConkey KM (1992). Reports of real and false memories: the relevance of hypnosis, hypnotizability, and context of memory test. J Abnorm Psychol 101:521– 527. Bartlett FC (1932). Remembering: A Study in Experimental and Social Psychology. Cambridge, England: Cambridge University Press. Bartlett FC (1983). Remembering, Cambridge, England: Cambridge University Press. Beck AT (1967). Depression: Clinical, Experimental, and Theoretical Aspects. New York: Harper & Row. Beck AT (1976). Cognitive Therapy and the Emotional Disorders. New York: International Universities Press. Beck AT, Rush AJ, Shaw BF (1979). Cognitive Therapy for Depression. New York: Guilford. Bekerian DA, Goodrich SJ (1995). Telling the truth in the recovered memory debate. Conscious Cogn 4:120–124. Blume ES (1990). Secret Survivors: Uncovering Incest and Its Aftereffects in Women. New York: Ballantine. Bonnano GA (1990). Remembering and psychotherapy. Psychotherapy 27:175–186. Brewer WF (1988). Memory for randomly sampled autobiographical events. In: Remembering Reconsidered: Ecological and Traditional Approaches to the Study of Memory, Neisser U, Winograd E, eds. Cambridge, England: Cambridge University Press. Buckner RL (1996). Beyond HERA: contributions of specific prefrontal brain areas to long-term memory retrieval, Psychon Bull Rev 3:149–158. Christianson S, Safer MA (1996). Emotional events and emotions in autobiographical memories. In: Remembering Our Past: Studies in Autobiographical Memory, Rubin DC, ed. Cambridge, England: Cambridge University Press, pp 218–243. Cialdini R (1993). Influence: Science and Practice, 3rd ed. Glenview, IL: Harper Collins. Cohler BJ (1994). Memory recovery and the use of the past: a commentary on Lindsay and Read from psychoanalytic perspectives. J Appl Cognit Psychol 8:365–378.
Annual progress in child psychiatry and child development 2000–2001
284
Courtois CA (1991). Theory, sequencing, and strategy in treating adult survivors. New Dir Ment Health Serv 51:47–60. Daley DC, Zuckoff A (1998). Improving compliance with the initial out-patient session among discharged inpatient dual diagnosis clients. Soc Work 43:470–473. Freud S (1899). Screen memories. In: Sigmund Freud: Collected Papers, Vol 5, Strachey J, ed. London: Hogarth Press, 1953, pp 47–69. Garry M, Loftus EF, Brown SW (1994). Memory: a river runs through it. Conscious Cogn 3:438–451. Goldstein E, Farmer K (1992). Confabulations: Creating False Memories, Boca Raton, FL: SIRS Books. Harvey MR, Herman JL (1994). Amnesia, partial amnesia, and delayed recall among adult survivors of childhood trauma. Conscious Cogn 3:295–306. Johnson MK, Hashtroudi S, Lindsay DS (1993). Source monitoring. Psychol Bull 114:3– 28. Johnson MK, Raye CL (1981). Reality monitoring. Psychol Rev 88:67–85. Kapur N (1993). Focal retrograde amnesia in neurological disease: a critical review. Cortex 29:217–234. Kihlstrom JF (1994). Hypnosis, delayed recall, and the principles of memory. Int J Clin Exp Hypn 42:337–345. Krass J, Kinoshita S, McConkey KM (1988). Hypnotic memory and confident reporting. J Appl Cognit Psychol 2:35–51. Lindsay DS, Read JD (1994). Psychotherapy and memories of childhood abuse. J Appl Cognit Psychol 8:283–338. Linton M (1978). Real world memory after six years: an in vivo study of very long term memory. In: Practical Aspects of Memory, Grunneberg MM, Morris PE, Sykes RN, eds. New York: Academic Press, pp 69–76. Linton M (1982). Transformations of memory in everyday life. In: Memory Observed: Remembering in Natural Contexts, Neisser U, ed. San Francisco: Freeman, pp 77–91. Loftus EF (1993). The reality of repressed memories. Am Psychol 48:518–537. Loftus EF, Loftus OR (1980). On the permanence of stored information in the human brain. Am Psychol 35:409–420. Lynn SI, Payne DO (1997). Memory as the theater of the past: the psychology of false memories. Curr Dir Psychol Sci 6:55. Maltz W (1990). Adult survivors of incest: how to help them overcome the trauma. Med Aspects Hum Sex 24:42–47. McConkey KM, Sheehan PW (1995). Hypnosis, Memory, and Behavior in Criminal Investigation. New York: Guilford. McGovern P, Lurie N, Margolis K, Slater S (1998). Accuracy of self-report of mammography and pap smear in a low-income urban population. Am J Prev Med 14:201–208. Mishkin M, Malamut B, Bachevalier I (1984). Memories and habits: two neural systems. In: Neuropsychology of Memory, Lynch O, McGaugh L, Weinberger NM, eds. New York: Guilford. Mitchell CO (1998). Perceptions of empathy and client satisfaction with managed behavioral health care. Soc Work 43:404–411.
The altering of reported experiences
285
Neisser U (1982). Memory Observed: Remembering in Natural Contexts. San Francisco: Freeman. Neisser U, Fivush R (1994). The Remembering Self: Construction and Accuracy in the Self Narrative. Cambridge, England: Cambridge University Press. Nyberg L, Cabeza R, Tulving E (1996). PET studies of encoding and retrieval: the HERA model. Psychol Bull Rev 3:135–148. Offer D (1969). The Psychological World of the Teenager. New York: Basic Books. Offer D, Offer JB (1975). From Teenage to Young Manhood: A Psychological Study. New York: Basic Books. Offer D, Ostrov E, Howard KI, Dolan S (1992). Offer Self-Image Questionnaire, Revised. Los Angeles: Western Psychological Services. Olio KA (1989). Memory retrieval in the treatment of adult survivors of sexual abuse. Transact Anal J 19:93–100. Petri HL, Mishkin M (1994). Behaviorism, cognitivism and the neuropsychology of memory. Am Sci 82:3–37. Pezdek K (1994). The illusion of illusory memory. J Appl Cognit Psychol 8:339–350. Pezdek K, Roe C (1994). Memory for childhood events: how suggestible is it? Conscious Cogn 3:374–387. Roediger HL III (1980). Memory metaphors in cognitive psychology. Mem Cognit 8:231–246. Roediger HL III (1990). Implicit memory: retention without remembering. Am Psychol 45:1043–1056. Rubin D, Schulkind M (1997). The distribution of autobiographical memories across the lifespan. Mem Cognit 25:859–866. Rubin DC, ed (1996). Remembering Our Past: Studies in Autobiographical Memory. Cambridge, England: Cambridge University Press. Schacter DL (1994). Priming and multiple memory systems: perceptual mechanisms of implicit memory. In: Memory Systems, Schacter DL, Tulving B, eds. Cambridge, MA: MIT Press. Smith ME (1952). Childhood memories compared with those in adult life. J Genet Psychol 80:151–182. Spence DP (1984). Narrative Truth and Historical Truth. New York: Norton. Spence DP (1994). Narrative truth and putative child abuse. Int J Clin Exp Hypn 42:289– 303. Squire LR (1987). Memory and Brain. New York: Oxford University Press. Squire LR (1992). Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans. Psychol Rev 99:195–231. Thompson CP, Skowronski JJ, Larsen SF, Betz AL (1996). Autobiographical Memory: Remembering What and Remembering When. Hillsdale, NJ: Erlbaum. Tulving E (1989). Remembering and knowing the past. Am Sci 77:361–367. Wagenaar WA (1986). My memory: a study of autobiographical memory over six years. Cognit Psychol 18:225–252. White RT (1982). Memory for personal events. Hum Learn 1:171–183. Woodruff DS, Birren JE (1972). Age changes in cohort differences in personality. Dev Psychol 6:252–259.
Annual progress in child psychiatry and child development 2000–2001
286
Wright J, Morley S (1995). Autobiographical memory and chronic pain. Br J Clin Psychol 34:255–265. Yapko M (1994). Suggestions of Abuse. New York: Simon & Schuster. Yarrow MR, Campbell JD, Burton RV (1970). Recollections of childhood: a study of the retrospective method. Monogr Soc Res Child Dev 35:1–3.
PART IV: OTHER CLINICAL ISSUES
15 Developmental Coordination Disorder in Swedish 7-Year-Old Children Björn Kadesjö and Christopher Gillberg
Objective: To estimate the prevalence, comorbidity, and outcome in developmental coordination disorder (DCD). Method: In this population study of 7-year-olds undergoing individual examination plus teacher and parent interviews, children were followed up at ages 8, 9, and 10 years. Results: Severe DCD occurred in 4.9% and moderate DCD in another 8.6%. Boy-girl ratios ranged from 4:1 to 7:1. Children with severe and moderate DCD did not differ from each other on any measure, but both groups were clearly separated from children without DCD with respect to associated attention deficit symptoms. Asperger disorder symptoms, school dysfunction scores, and outcome. Approximately half of all children with DCD had moderate to severe symptoms of attention deficit-hyperactivity disorder (ADHD). Conclusions: DCD is a common problem, and it is strongly associated with ADHD symptoms. A diagnosis of DCD at age 7 years predicts DCD at age 8 years and restricted reading comprehension at age 10 years. Clinicians need to acquaint themselves with DCD and its comorbidity so that they can provide better services to affected children. (J Am Acad Child Adolesc Psychiatry, 1999, 38(7):820– 828.) Key Words: developmental coordination disorder, clumsiness, attention deficit-hyper activity disorder, Asperger’s disorder, reading disorder.
INTRODUCTION Some children lack the motor skills required for everyday activities such as play, sports, and schoolwork. Although not generally delayed, they usually do not have an easily identifiable neurological disorder. Their motor difficulties are clinically significant, regardless of whether or not they should be interpreted as a sign of neurological disorder. The DSM-III-R (American Psychiatric Association, 1987) defined “developmental coordination disorder” (DCD) as motor coordination performance markedly inappropriate for age and IQ causing significant interference with academic achievement or activities of daily living. DCD is listed in the DSM-IV (American Psychiatric Association, 1994) as a
Developmental coordination disorder
289
specific disorder of development. In earlier literature it has been variously referred to as “clumsy child syndrome” (Gordon & McKinlay, 1980), “motorperception dysfunction” (Gillberg et al., 1982), or “minor neurological dysfunction” (Hadders-Algra & Touwen, 1992). Related problems have been dealt with in the considerable literature on “soft neurological signs” (Reeves & Werry, 1987; Rie, 1987; Tupper, 1987). Several studies have suggested that such signs are nonspecific indicators of behavioral/emotional problems such as schizophrenia, anxiety, depression, and obsessive-compulsive disorder (Malla et al., 1997; Pine et al., 1993, 1997). The inclusion of DCD in a manual of psychiatric disorders suggests that it is associated with behavior problems, but there has been little study in this field. Motor clumsiness is correlated with cognitive and perceptual problems (Wilson & McKenzie, 1998), but it is not known whether DCD or perceptual problems are correlated with psychiatric disorder or with other kinds of problems, such as reading disorder, as has been suggested by several authors (e.g., Hellgren et al., 1994). The DSM definitions of DCD do not provide clear cutoffs vis-à-vis normality. Children’s environments vary with regard to demands and expectations of motor performance. Tradition and culture determine children’s experience with motor activities. DCD is often stated to occur at a rate of approximately 5% of children, but Henderson and Sugden (1992) believe that another 10% have similar but milder problems. “Poor coordination” was found in 8.1% of approximately 30,000 7-year-olds followed in the Perinatal Collaborative Project (Nichols, 1987). In another study of 1,443 schoolchildren, the rates varied from 5.4% at 6 years to 1.3% at 10 years (van Dellen et al., 1990). Whitmore and Bax (1990) followed up 5-year-old children for 2 and 5 years. They found that 25% and 46% of children with deviant neurodevelopmental scores at age 5 years had learning disability or behavior problems at follow-up (compared with 4% and 8% of children without abnormal neurodevelopmental scores). Clumsy children may be more introverted and have less self-confidence with respect to physical and social skills (Schoemaker & Kalverboer, 1990), have feelings of inferiority (Gordon & McKinlay, 1980), and be less well-liked by peers (Gubbay, 1975). Increased rates of behavior problems, affective disorders, school adjustment difficulties, and other social problems have also been reported (Cantell et al., 1994; Gillberg & Gillberg, 1989b; Gueze, 1993; Gueze & Börger, 1993; Losse et al., 1991; Michelsson & Lindahl, 1993). Gillberg and Gillberg (1989b) reported that 65% of children with motorperception dysfunction have “attention deficit disorder.” Several studies by Denckla (Denckla, 1973; Denckla & Rudel, 1978; Denckla et al., 1985), Wolff (1990), Gillberg (Gillberg et al., 1982, 1983; Kadesjö & Gillberg, 1998; Landgren et al., 1996), and other authors (Szatmari et al., 1989b; Witmont & Clark, 1996) have indicated that there is a strong relationship between attention deficit-hyperactivity disorder (ADHD) and motorperception dysfunction. The combination of DCD and ADHD is sometimes referred to as DAMP (deficits in attention, motor control, and perception), a terminology that is in widespread use in the Nordic countries (Airaksinen et al., 1991). According to population studies performed in many centers in Scandinavia (e.g., Gillberg, 1985; Hellgren et al., 1993, 1994; Kadesjö & Gillberg, 1998), DAMP may have stronger validity in terms of common background factors and poorer psychosocial/academic outcome than either ADHD or DCD. ADHD
Annual progress in child psychiatry and child development 2000–2001
290
and DCD might be preferred diagnostic terms for reasons of “purity,” but there is widespread realization that DCD is so often associated with ADHD symptoms that a term acknowledging both types of deficit is needed in clinical practice. This study examines DCD: (a) prevalence in a total population sample of 7year-olds; (b) diagnostic boundaries; (c) impact in terms of its relationship to attention deficits and other behavior problems; and (d) predictive power with respect to later speech-language and reading problems. The underlying hypotheses are that: (a) DCD is a common problem affecting a number of young school-age children; (b) DCD is strongly associated with attention deficits and to some extent with other behavior problems; (c) DCD is a marker for speechlanguage and reading disorder; and (d) “moderate DCD” is a “gray zone” between normal motor functioning and severe DCD, associated behavior problems and future language/reading disorders being more marked in both DCD groups, but significantly more so in the severely affected group.
METHOD The study involves exactly half of all children born in 1985 and attending normal schools (first grade) in the community of Karlstad in the autumn of 1992. Swedish children start school at age 7 years. Karlstad is located in central Sweden (population 70,000) and has a social class distribution, immigrant population, level of income, and education corresponding to that of Sweden as a whole (Statistics Sweden, 1998; Swedish Association of Local Authorities, 1997). Population Sample There were 818 children (433 boys, 385 girls, representing 52.9% and 47.1 %, respectively) born in 1985 and attending normal classrooms in the autumn term of 1992. Eight additional children (five boys and three girls)—or 1%—were in mental retardation classrooms. They had been comprehensively neurodevelopmentally assessed before the present study. They had autistic disorder (three boys and one girl who also had cerebral palsy-ataxia), Down syndrome (one boy and one girl), and cerebral palsy-hemiplegia (one boy also had ataxia and one girl also had epilepsy). These 826 children constituted the total population born in 1985 and living in Karlstad in 1992. Half of the 818 children attending normal classrooms were assigned for study (224 boys, 185 girls, representing 54.8% and 45.2%, respectively). They attended 12 of the 25 schools in Karlstad. This 50% sample is considered representative of Karlstad children attending normal classrooms. Twenty-one of the 409 children (5.1%) had at least one parent who had migrated from outside the Nordic countries. Study Group and Procedure The children were 6 years 8 months through 7 years 8 months at initial evaluation (time 0m). The physical education teacher evaluated each child’s motor performance during the first month of the autumn term and again 8 months later (time 8m). Children were
Developmental coordination disorder
291
examined individually, and at least one parent was interviewed by the first author in the autumn of 1992 (time 4m). There was no attrition. Examination and interview lasted 15 to 45 minutes, longer if the child had major problems. For each child, the grade teacher and special education teacher completed a screening questionnaire, and both were interviewed by the first author about behavior and academic performance in the classroom (time 4m). The teachers then completed the brief Conners scale (Conners, 1969) (time 8m). The study was performed within the context of a school health setting requiring the use of simple, brief, and easy-to-administer assessment measures. Many published tests were considered too cumbersome for practical application in this situation. Most of the tests used in the study have acceptable to excellent psychometric properties, but reliability data were lacking for some. In such instances, psychometrically validated tests tapping similar functions were added. Thus, for instance, psychometric data pertaining to the teacher motor examination (see next section) were lacking when the study was launched, but testretest reliability was subsequently demonstrated to be acceptable to good (see “Results” section), and findings were validated against the medical motor examination, which has good to excellent psychometric properties. Follow-up was accomplished approximately 20 monthes (time 20m) and 32 months (time 32m) after the initial study. A test of reading comprehension (see the following) was given at time 20m and time 32m. At time 20m, a part of the Illinois Test of Psycholinguistic Abilities (see the following) was given. Methods Used Motor Examinations of the Child (Time 0m, Time 4m, and Time 8m). The physical education teacher examined approximately eight children at a time, using portions of the Folke Bernadotte test (Bille et al., 1985), including 11 motor items (jumping with feet together, hopping on left and right foot, alternating jumping left-right foot forward, walking with toes turned outward and inward, throwing and catching a ball, tying a knot, alternating movement—right hand to left chest, left hand to right chest, standing with arms outstretched while performing pronation-supination movements, and diadochokinesis). No systematic psychometric data exist for this test. Each session was videotaped, and the teacher scored each of the 11 items as 0 (no abnormality), 1 (some abnormality), or 2 (major abnormality). The teacher motor dysfunction score (time 0m) ranged from 0 through 22. At time 4m, the first author examined the motor status of each child by using a modified version of the test developed by Gillberg et al. (1983), comprising 11 gross and fine motor items (hopping on left and right foot, standing on left and right foot, diadochokinesis left and right, the Fog test, jumping back and forth across a line, alternating jumping left-right foot forward, finger tapping, and finger imitation). Each item was scored according to the same scale as was used by the physical education teacher. The medical motor dysfunction score (time 4m) ranged from 0 through 22. This score formed the basis for the diagnostic classification of DCD (see the following). The first 7 of these items have excellent interrater reliability (Gillberg, 1985) and are highly valid in discriminating between children identified as clumsy versus those identified as
Annual progress in child psychiatry and child development 2000–2001
292
nonclumsy after detailed neuromotor examination with a comprehensive battery (Gillberg et al., 1983). Eight months later, the children were reevaluated by another physical education teacher using the same procedure as at time 0m. A teacher motor dysfunction score (time 8m) was computed for each child. At this examination the children ranged in age from 7 years 4 months to 8 years 4 months. Child Behavior Observation (Time 4m). The child’s behavior was observed at parent interview and at pediatric/motor examination. The first author scored: (i) attention deficits; (ii) hyperactivity; (iii) impulsivity; or (iv) distractibility, separately as no problems (0), some problems (1), or major problems (2). An observed hyperactivity score (time 4m) was computed by adding (i) through (iv). The range was 0 through 8. Parent Interview (Time 4m). At least one parent of each child received a semistructured interview focusing on the child’s developmental milestones, attention, social interaction, behavior in general, motor performance, and problems in any of these domains. Previous or current illnesses were noted. The parent was asked whether the child had difficulties attending to tasks and, if so, whether these interfered with overall functioning. Only if they did was attention deficit in the home (time 4m) diagnosed. Phonological Development, Letter Knowledge, and Reading Ability (Time 0m). Phonological development was assessed by the teacher, using a phoneme segmentation test (Lundberg et al., 1984) at time 0m, scored as 0 (no problems), 1 (some problems), and 2 (major problems). The ability to read aloud was assessed by this teacher and scored according to the same scale. The number of letters known (possible range 0 through 28) by the child on starting school was individually ascertained by a special education teacher. Teacher Questionnaire Screening (Time 4m). Each child had a teacher questionnaire completed during the fall term of 1992. The questionnaire covered five areas: (i) attention (regardless of activity level of child); (ii) social interaction (children and adults); (iii) learning (in general terms and specifically with regard to reading and writing); (iv) language (comprehension and expressive skills); and (v) emotion (anxiety/depression or aggressive/acting-out behaviors). Scores of 0, 1, and 2 were awarded as described above. All area scores were added to yield a school dysfunction score (time 4m) ranging from 0 through 10. Teacher Interview (Time 4m). The teacher was interviewed by the first author about possible problems in the 5 domains covered by the teacher questionnaire. For each child with a score of 1 or 2 in any questionnaire domain, he probed into the specific problems shown by the child. For children scoring 1 or 2 in the attention domain on the questionnaire or if teachers mentioned attention deficits during the interview, a DSM-IIIR (American Psychiatric Association, 1987) checklist for ADHD—with excellent interrater reliability (Landgren et al., 1996)—was used in the continued teacher interview. The 14 ADHD items were covered in detail and were endorsed only if pervasive and considered to impinge on the child’s functioning in several different settings. For each child scoring 1 or 2 in the social interaction domain or if the teacher mentioned social interaction problems during the interview, checklists covering Asperger’s disorder criteria published by Gillberg and Gillberg (Ehlers & Gillberg, 1993; Gillberg & Gillberg, 1989a) and Szatmari (1989a) were used in the continued teacher
Developmental coordination disorder
293
interview. Both checklists have good to excellent interrater reliability. For each child scoring 1 or 2 in the emotional domain or whose teacher mentioned aggressive behaviors or peer relationship problems during the interview, a DSM-III-R (American Psychiatric Association, 1987) checklist for oppositional defiant disorder (ODD)—not tested for reliability—was used by the first author in the continued teacher interview. Teacher Conners Follow-up Screening (Time 8m). When each child had spent approximately 8 months in school, the teacher completed the short version of the Conners teacher scale (Conners, 1969). This scale comprises 10 attentionactivity leveloppositional defiant items that are scored on a 0, 1, 2, 3 scale. The possible Conners follow-up score (time 8m) ranged from 0 through 30, with higher scores indicating more problems. Illinois Test of Psycholinguistic Abilities Follow-up (Time 20m). All children were given a part of the Illinois Test of Psycholinguistic Abilities (Kirk et al., 1968) (auditory closure, auditory association, grammatic closure, auditory sequential memory, visual sequential memory) approximately 20 months after initial diagnostic evaluation. Each subscale was scored 1 through 9, with 9 indicating the highest abilities. The possible range of scores for the summed subscale results was 5 through 45. Reading Comprehension Follow-up (Time 20m and Time 32m). A modified Swedish test of reading comprehension (Björkquist & Järpsten, 1975, 1983) was administered individually to all children approximately 20 and 32 months after school began (possible range of score 0 through 32). Diagnostic Classification Neurodevelopmental/neuropsychiatric classifications were made by the first author on the basis of the results obtained at motor examination of the child and teacher interview. Children receiving a diagnosis were all considered to be impaired by the diagnostic symptoms in daily life activities (including overall school functioning). Moderate DCD. The criterion level for this category was a medical motor dysfunction score of 10 to 14. The reason for cutoff at a motor dysfunction score of 10 was that there was a sharp decline in the number of children with scores of 10 compared with 9 and then a second “hump” of high scorers (see Figure 15.1). Severe DCD. The criterion level for this category was a medical motor dysfunction score of 15 or greater. Severe ADHD. The criterion level for this category was 8 or more of the 14 DSM-III-R ADHD criteria ascertained at teacher interview. Moderate ADHD. The criterion level for this category was 5, 6, or 7 of the 14 DSMIII-R ADHD criteria. The reason for cutoff at five symptoms was that all except one (2%) of the 54 children reported by the teacher to have major attention problems met five or more ADHD symptom criteria, whereas only four (1%) of the 355 children with no or only some attention problems did. Mainly Inattention/Mainly Hyperactivity. The 14 DSM-III-R ADHD symptoms were designated as either inattention markers (items A3, A6, A7, A12, and A13) or hyperactivity markers (items A1, A2, A4, A5, A8, A9, A10, A11, and A14).
Annual progress in child psychiatry and child development 2000–2001
294
Figure 15.1 Medical motor dysfunction score (time 4m). DCD=developmental coordination disorder. Normal. Children not meeting criteria for DCD or ADHD were classified as normal for the purpose of the present data analysis. Some of these children would meet criteria for non-ADHD psychiatric disorders. “Nonclumsiness” refers to scores of 9 or less on the medical motor dysfunction score. Statistical Analyses Ninety-five percent confidence intervals (Cls) were calculated whenever appropriate. The χ2 test with the Yates correction was used in comparing group frequencies. One-way analysis of variance was used for overall comparison of group means. When multiple comparisons were made, the modified least significant difference test of Bonferroni was used.
RESULTS DCD Prevalence Twenty children (4.9% of the population)—18 boys, two girls—had severe DCD, and 35 (8.6% of the population)—29 boys, six girls—had moderate DCD. The rate of severe DCD among boys in the population was 8.0%, and among girls it was 1.1% (male-female 7.3:1). Corresponding rates for moderate DCD were 12.9% and 3.2% (male-female 4:1). In addition, two boys were judged to have specific neurological disorders (cerebral palsy, surgically corrected spina bifida with myelomeningocele). The children with DCD were similar to those without DCD in terms of immigrant status and parents’ occupation.
Developmental coordination disorder
295
Correspondence Between Physician and Teacher Judgment of Clumsiness Agreement between medical motor dysfunction scores at time 4m and teacher motor dysfunction scores at time 0m and at time 8m was excellent (r=0.84 and 0.72, respectively; p<.001 in both instances). Teacher dysfunction scores at time 0m and time 8m correlated strongly with each other (r=0.74, p< .001). Physician and teacher (time 0m) agreed about nonclumsiness in 81% of the cases. Of 20 individuals with severe DCD diagnosed by the physician, 19 (95%) were considered clumsy (severe or moderate) at teacher examination (time 0m). Of 53 individuals with DCD (severe or moderate) diagnosed by the physician, 42 (79%) were considered clumsy (severe or moderate) by the teacher. Of 21 children considered severely clumsy by the teacher (time 0m), 20 (95%) received a physician diagnosis of severe or moderate DCD. Of 63 individuals considered clumsy by the teacher (severe or moderate), 42 (67%) received a physician diagnosis of severe or moderate DCD. Relationship of DCD to ADHD Children with DCD showed attention deficit in the home more often than the group without this diagnosis (Table 15.1), and they had higher mean observed hyperactivity scores at the medical examination (time 4m). DCD (severe and moderate) correlated strongly with ADHD diagnosis and with number of ADHD symptoms. Of children with DCD, 47% had five or more ADHD symptoms. Full criteria for severe ADHD were met by 19% of the DCD group. Of those with severe DCD, 55% had five or more ADHD symptoms and 15% met full criteria for severe ADHD. Corresponding figures for 354 children not meeting criteria for DCD were 9% and 2%, respectively (p<0.001 in both instances). Mean number of ADHD criteria was 1.0 (CI 0.8, 1.2) in the non-DCD group, 4.2 (CI 3.4, 5.0) in the moderate, and 4.7 (CI 3.3, 6.1) in the severe DCD group. The DCD groups did not differ from each other, but both differed from the non-DCD group (p<0.01). Mean number of inattention criteria in the groups was 0.5 (CI 0.3,0.6), 2.3 (CI 1.7, 2.9), and 3.0 (CI 2.2, 3.8) (p<×.01) (maximum possible 5.0). Mean number of hyperactivity criteria in each group was 0.5 (CI 0.4, 0.6), 1.9 (CI 1.4, 2.4), and 1.7 (CI 0.8, 2.6) (p<×.01) (maximum possible 9.0). Conners follow-up scores were higher in the DCD group than in the non-DCD group. There was a marked interaction of DCD and ADHD in predicting number of teacherreported ADHD symptoms, attention deficit in the home, and Conners follow-up score (Table 15.2). There were independent main effects of a diagnosis of DCD and, obviously, of ADHD on various measures of attention deficit severity.
Annual progress in child psychiatry and child development 2000–2001
296
TABLE 15.1 DCD and Associated Attention/Activity Problems
No. of ADHD Deficit in the Criteria Home Home (Time 4m) (Time 4m) (Range 0–1) (Range 0–1)
Attention Hyperactivity Exam Exam (Time 4m) (Range 0–2)
Observed Conners Teacher Follow-up Score (Time 8m) (Range 0–30)
DCD at Age Mean SD Mean SD Mean SD Mean SD 7 Years DCD All 4.4 2.9 0.3 0.2 1.8 1.8 8.2 6.8 Severe 4.7 3.3 0.4 0.5 2.1 1.6 7.2 5.3 Moderate 4.2 2.6 0.3 0.5 1.7 1.6 8.9 7.5 With 6.9 1.6 0.7 0.5 2.3 2.0 12.8 6.3 ADHD No ADHD 2.0 1.4 0.0 0.0 1.3 1.5 3.8 3.5 No DCD All 1.0 2.1** 0.1 0.2** 0.3 0.8** 3.3 5.5** With 6.8 1.5 0.5 1.8 1.5 1.8 17.0 6.2 ADHD No ADHD 0.4 1.0* 0.0 0.0† 0.2 0.5* 2.0 3.2* Note. DCD=developmental coordination disorder; ADHD=attention deficithyperactivity disorder. †p<.05 no DCD no ADHD versus DCD with ADHD and no DCD with ADHD; *p<.05 no DCD no ADHD versus all other groups (no DCD with ADHD, DCD with ADHD, DCD no ADHD); **p<.01 no DCD versus DCD (DCD all, DCD severe, and DCD moderate). TABLE 15.2 Contribution of DCD and ADHD, Separately and Combined, to Behavior and Reading/Language Competence
No. of ADHD criteria, teacher interview Attention deficit in the home Observed hyperactivity, medical exam Conners teacher follow-up score
ADHD F1=1072, 9 p<0.001 F1=275, 9 p<0.001 F1=83, 5 p<0.001 F1=482, 5 p<0.001
DCD F1=420, 9 p<0.001 F1=125, 4 p<0.001 F1=25, 5 p<0.001 F1=80, 5 p<0.001
ADHD+DCD F1=17, 4 p<0.001 F1=7, 9 p<0.005 F1=0, 2 NS F1=22, 5 p<0.001
Developmental coordination disorder
297
ODD symptoms
F1=105, 6 F1=12, 6 F1=1, 9 p<0.001 p<0.001 NS a F =120, 5 F =217, 0 F =53, 5 Asperger’s symptoms, Gillberg 1 1 1 p<0.001 p<0.001 p<0.001 b F =295, 8 F =388, 5 F Asperger’s symptoms, Szatmari 1 1 1=50, 1 p<0.001 p<0.001 p<0.001 F1=31, 1 F1=0, 2 Letter knowledge F1=6, 6 p<0.05 p<0.001 NS F1=21, 2 F1=0, 3 Reading aloud F1=1, 8 NS p<0.001 NS F1=31, 0 F1=1, 0 Segmentation F1=6, 8 p<0.01 p<0.001 NS F1=29, 2 F1=2, 3 ITPA (part) F1=7, 1 p<0.01 p<0.001 NS F1=21, 8 F1=0, 5 Reading comprehension (time F1=12, 1 20m) p<0.001 p<0.001 NS F1=45, 7 F1=0, 7 Reading comprehension (time F1=15, 2 32m) p<0.001 p<0.001 NS Note: ADHD=moderate and severe attention-deficit/hyperactivity disorder; DCD=moderate and severe developmental coordination disorder; ADHD+DCD=criteria for ADHD and DCD; ODD=oppositional defiant disorder; ITPA=Illinois Test of Psycholinguistic Abilities; NS= not significant. a Ehlers and Gillberg, 1993; Gillberg and Gillberg, 1989a. b Szatmari et al., 1989a. School Dysfunction Mean school dysfunction score was 1.1 (CI 0.9,1.3) in the non-DCD group, 4.3 (CI 3.4, 5.2) in the moderate, and 5.0 (CI 4.0, 6.0) in the severe DCD group (p<0.001). The two DCD groups did not differ from each other. DCD children with ADHD (moderate or severe) had mean scores of 6.1 (CI 5.3, 6.9), whereas those without ADHD had mean scores of 3.1 (CI 2.3, 3.9) (p<0.001). NonDCD children with ADHD had mean scores of 4.5 (CI 3.8, 5.2), whereas those without ADHD had mean scores of 0.8 (CI 0.7, 0.9) (p<0.001). The school dysfunction score included a specific measure of attention deficit. Relationship of DCD to ODD Both DCD and ODD were strongly comorbid with ADHD. Children with severe ADHD with and without DCD had means of 3.8 and 2.7 ODD symptoms, respectively (not significant). Those with moderate ADHD with and without DCD had corresponding means of 1.1 and 1.0. The mean in children with DCD without ADHD was 0.1, similar to
Annual progress in child psychiatry and child development 2000–2001
298
normal children (0.1, not significant). Relationship of DCD to Asperger’s Disorder Symptoms Children with DCD had more symptoms of Asperger’s disorder than children without DCD (Table 15.3). Four children met full criterial for Asperger’s disorder. ADHD, independently of DCD, predicted Asperger’s disorder symptoms, but DCD-ADHD combined had an interactive effect and a very strong correlation with such symptoms. DCD as a Predictor of Speech-Language and Reading Problems Children with DCD scored worse on tests of speech, phonology, and reading (Table 15.4). Reading comprehension at 9 and 10 years was compromised (Table 15.5). The combination of DCD and ADHD tended to yield the lowest reading comprehension scores, both at age 9 and 10 years, but there was no statistically significant interaction between DCD and ADHD (Table 15.2).
TABLE 15.3 DCD and Associated Symptoms of ODD and Asperger’s Disorder
Asperger’s Symptoms, Asperger’s Symptoms, ODD Symptoms Gillberga Szatmarib (Range 0–20) (Range 0–22) (Range 0–7) Mean SD Mean SD Mean SD
DCD at Age 7 Years DCD All 0.7 1.4 2.3 3.7 3.2 4.9 Severe 0.5 1.1 2.9 3.7 4.1 5.2 Moderate 0.9 1.6 2.0 3.7 2.7 4.7 With ADHD 1.4 1.8 4.0 4.3 5.4 5.7 No ADHD 0.1 0.6 0.7 1.7 1.0 2.4 No DCD All 0.2 1.0** 0.2 0.8** 0.3 1.4** With ADHD 1.8 2.4 1.0 1.6 2.0 3.2 No ADHD 0.1 0.6* 0.1 0.6* 0.2 1.0* Note. ADHD=attention deficit-hyperactivity disorder; DCD=developmental coordination disorder; ODD=oppositional defiant disorder. a Ehlers & Gillberg, 1993; Gillberg & Gillberg, 1989a. b Szatmari et al., 1989a. *p<0.05 no DCD no ADHD versus no DCD with ADHD and no DCD no ADHD versus DCD with ADHD; **p<0.01 no DCD versus DCD all.
Developmental coordination disorder
299
TABLE 15.4 DCD as a Predictor of Language Abilities
Letter Knowledge Reading Aloud Segmentation ITPA Subtests (Range 0–2) (Range 5–45) (Range 0–28) (Range 0–2) (Time 0m) (Time 0m) (Time 20m) (Time 0m) Mean SD Mean SD Mean SD Mean SD
DCD at Age 7 Years DCD All 14.6 9.0 0.1 0.5 0.6 0.8 18.8 5.6 Severe 16.0 9.5 0.1 0.4 0.5 0.8 18.5 6.1 Moderate 13.7 8.7 0.2 0.6 0.6 0.9 18.9 5.4 With ADHD 13.3 8.9 0.1 0.4 0.4 0.6 18.6 4.8 No ADHD 16.0 9.0 0.2 0.6 0.8 1.0 19.0 6.3 No DCD All 21.4 8.0** 0.8 0.9** 1.3 0.9** 23.3 5.8** With ADHD 17.7 8.2 0.6 0.9 1.0 0.9 20.3 5.9 No ADHD 21.8 7.9* 0.9 0.9* 1.3 0.8* 23.6 5.7* Note. ADHD=attention deficit-hyperactivity disorder; DCD=developmental coordination disorder; ITPA=Illinois Test of Psycholinguistic Abilities. *p<0.05 no DCD no ADHD versus DCD with ADHD, DCD no ADHD; **p<0.01 no DCD versus DCD (DCD all, DCD severe, and DCD moderate). TABLE 15.5 DCD as a Predictor of Later Reading Comprehension
Reading Comprehension Time 20m Time 32m Mean SD Mean SD
Increase in Score Mean
SD
DCD All 4.1 3.8 8.2 4.9 4.1 3.7 Severe 3.2 3.9 7.9 5.7 4.7 3.8 Moderate 4.6 3.8 8.3 4.5 3.7 3.6 With ADHD 3.2 3.0 7.0 4.7 3.8 3.3 No ADHD 4.9 4.3 9.3 4.9 4.4 3.9 No DCD All 7.0 4.4** 13.8 5.5** 6.8 3.8** With ADHD 4.5 3.5 10.3 5.0 5.8 3.7 No ADHD 7.3 4.4* 14.2 5.4* 6.9 3.9† Note. ADHD=attention deficit-hyperactivity disorder; DCD=developmental coordination disorder. †p<0.05 no DCD no ADHD versus DCD with ADHD and no DCD with ADHD;
Annual progress in child psychiatry and child development 2000–2001
300
*p<0.05 no DCD no ADHD versus all other groups (no DCD with ADHD, DCD with ADHD, DCD no ADHD); **p<0.01 no DCD versus DCD (DCD all, DCD severe, and DC moderate). Differences Between Moderate and Severe DCD Moderate and severe DCD were similar in terms of concomitant ADHD, ODD, Asperger symptoms, language problems, and reading disorder (see Tables). Moderate and severe DCD both differed significantly from non-DCD in all these respects.
DISCUSSION Our sample, although small, is likely to be representative of all Swedish 7year-olds. Karlstad is a typical Swedish community with a middle-sized town, suburbs, and some rural areas. Socioeconomics status is on par with that of Sweden as a whole. The small sample size is a drawback, but this is partly outweighed by the fact that each child was seen individually, diagnosis was based on personal examination, and there was no attrition. Most studies in the field have used two-stage epidemiological designs with primary screening and examination only of high-scoring children (e.g., Boyle et al., 1987; Gillberg et al., 1982). Such a design entails a risk of producing false-negatives at screening. That all children in the present population were personally examined minimizes the risk of producing false-negative cases. Another drawback is the lack of an IQ test at the first examination. However, all children were in normal classrooms and believed to be of normal intelligence. A test (Raven) was performed when the children were 11 years of age, and IQ was found to be weakly (albeit significantly) correlated with the medical motor dysfunction score (time 4m). The level of correlation (r=0.18) was lower than in the study by Shafer et al. (1986), who found r=0.30 when correlating “soft signs” with IQ. DCD is common; severe variants affect almost 5% of 7-year-olds, and another 8% or 9% have moderate problems. This is in accord with Wright and Sugden (1996), who found 4% DCD among 427 Singaporean primary school children. Our findings indicate that DCD shows stability over time and across raters. Children with moderate and severe DCD were similar with regard to comorbid attentional, language, and reading problems, and both groups differed markedly from those with no DCD. Thus, contrary to our original hypothesis, it would seem reasonable to include moderate DCD cases in a “clinically relevant” group, which would comprise approximately 13% of the general population of children. DCD was very often associated with ADHD symptoms. Approximately half of all children with DCD had concomitant ADHD (moderate or severe). The rate of comorbidity of ADHD and DCD (motor-perception dysfunction) was similar in another population-based study from Sweden, performed in the early 1990s, with roughly half of all 6-year-olds with motor-perception dysfunction meeting criteria for ADHD (Landgren et al., 1996). The diagnosis of DCD was based on findings at individual motor
Developmental coordination disorder
301
examination performed by a physician. The diagnosis of ADHD was based on findings obtained at a highly structured teacher interview (with excellent interrater reliability) covering the DSM-III-R ADHD criteria. Even though this interview was performed by the same physician, it was not immediately temporally connected with the motor examination and he did not have available the results of that assessment when he performed the interview. Thus, we believe that there was reasonable independence between the two diagnostic classifications. Children with DCD had high rates of ODD, but this type of problem was predicted better by the comorbidity with ADHD. Asperger’s disorder symptoms were predicted independently both by DCD and ADHD. There was an interaction effect of DCD and ADHD on Asperger’s symptoms and on the number of ADHD criteria met at original teacher interview, attention deficit in the home, and Conners follow-up scores. The combination of DCD and ADHD showed strong correlation with school dysfunction scores and later reading problems, even though there was no significant interactive effect of the two on these follow-up measures. Clinical Implications Motor clumsiness/DCD has been regarded as the territory of child neurologists and developmental pediatricians, whereas attention disorders are conceptualized as falling within the domains of child psychiatry. It is possibly this “split” that accounts for the fact that some psychiatrists are not aware of, much less appreciate, the implications of and sometimes the need to do something about the motor-perceptual problems that are so often comorbid with childhood ADHD. Conversely, child neurologists sometimes fail to appreciate the impact of attention deficits in the clumsy children referred to them for diagnosis and workup.
ACKNOWLEDGMENT Supported by the Swedish Medical Research Council (grant K97–21X11251–03CK).
REFERENCES Airaksinen E, Bille B, Carlström G et al. (1991). Barn och ungdomar med DAMP/ MBD (Children and adolescents with DAMP/MBD). Läkartidningen 88:714. American Psychiatric Association (1987). Diagnostic and Statistical Manual of Mental Disorders, 3rd edition-revised. Washington, DC: Author. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Bille B, Brieditis K, Steen M, Ekström B (1985). FBH-provet (in Swedish). Stockholm: Liber Utbildningsförlaget. Björkquist L-M, Järpsten B (1975, 1983). Diagnostiska läs-och skrivmaterial. (Diagnostic reading and writing material). Stockholm: Psykologiförlaget AB.
Annual progress in child psychiatry and child development 2000–2001
302
Boyle MH, Offord DR, Hofmann HG et al. (1987). Ontario Child Health Study, I: methodology. Arch Gen Psychiatry 44:826–831. Cantell MH, Smyth MH, Ahonen TP (1994). Clumsiness in adolescence: educational, motor and social outcomes of motor delay detected at 5 years. Adapted Physical Activity Quarterly 11:113–129. Conners CK (1969). A teacher rating scale for use in drug studies with children. Am J Psychiatry 126:884–888. Denckla M (1973). Development of coordination in normal children. Dev Med Child Neurol 16:729–741 Denckla M, Rudel R (1978). Anomalies of motor development in hyperactive boys. Ann Neurol 3:231–233. Denckla M, Rudel R, Chapman C, Krieger J (1985). Motor proficiency in dyslexic children with and without attentional disorders. Arch Neurol 42:228–231. Ehlers S, Gillberg C (1993). The epidemiology of Asperger syndrome: a total population study. J Child Psychol Psychiatry 34:1327–1350. Gillberg C, Carlström G, Rasmussen P (1983). Perceptual, motor and attentional deficits in Swedish primary school children: neurological screening aspects. Acta paediatr Scand 72:119–124. Gillberg C, Rasmussen P, Carlström G, Svensson B, Waldenström E (1982). Perceptual, motor and attentional deficits in six-year-old children: epidemiological aspects. J Child Psychol Psychiatry 23:131–144. Gillberg 1C (1985). Children with minor neurodevelopmental disorders, III: neurological and neurodevelopmental problems at age 10. Dev Med Child Neurol 27:3–16. Gillberg IC, Gillberg C (1989a). Asperger syndrome—some epidemiological considerations: a research note. J Child Psychol Psychiatry 30:631–638. Gillberg 1C, Gillberg C (1989b). Children with preschool minor neurodevelopmental disorders, IV: behaviour and school achievement at age 13. Dev Med Child Neurol 31:3–13. Gordon N, McKinlay I (1980). Helping Clumsy Children. New York: Churchill Livingstone. Gubbay S (1975). The Clumsy Child: A Study of Developmental Apraxia and Agnosic Ataxia. London: Saunders. Gueze R (1993). Longitudinal and Cross-Sectional Approaches in Experimental Studies in Motor Development. Cambridge, England: Cambridge University Press. Gueze R, Börger H (1993). Children who are clumsy: five years later. Adapted Physical Activity Quarterly 10:21. Hadders-Algra M, Touwen B (1992). Minor neurological dysfunction is more closely related to learning difficulties than to behavioral problems. J Learn Disabil 25:649– 657. Hellgren L, Gillberg C, Gillberg IC, Enerskog I (1993). Children with deficits in attention, motor control and perception (DAMP) almost grown up: general health at age 16 years. Dev Med Child Neurol 35:881–892. Hellgren L, Gillberg IC, Bagenholm A, Gillberg C (1994). Children with deficits in attention, motor control and perception (DAMP) almost grown up: psychiatric and personality disorders at age 16 years. J Child Psychol Psychiatry 35:1255–1271.
Developmental coordination disorder
303
Henderson S, Sugden D (1992). Movement Assessment Battery for Children: Manual San Antonio, TX: Psychological Corporation. Kadesjö B, Gillberg C (1998). Attention deficits and clumsiness in Swedish 7-yearold children. Dev Med Child Neurol 40:796–804. Kirk S, McCarthy J, Kirk W (1968). The Illinois Test of Psycholinguistic Abilities, rev ed. Urbana, IL: University of Illinois Press. Landgren M, Pettersson R, Kjellman B, Gillberg C (1996). ADHD, DAMP and other neurodevelopmental/neuropsychiatric disorders in six-year-old children: epidemiology and comorbidity. Dev Med Child Neurol 38:891–906. Losse A, Henderson SE, Elliman D, Hall D, Knight E, Jongmans M (1991). Clumsiness in children: do they grow out of it? A 10-year follow-up study. Dev Med Child Neurol 33:55–68. Lundberg I, Torneus M, Taube K (1984). UMESOL; Umeå skriv-och läs-material (in Swedish). Stockholm: Psykologiförlaget. Malla AK, Norman RM, Aguillar O, Cortese L, (1997). Relationship between neurological “soft signs” and syndromes of schizophrenia. Acta Psychiatr Scand 96:274–280. Michelsson K, Lindahl E (1993). Relationship between perinatal risk factors and motor development at the ages of 5 and 9 years. In: Motor Development in Early and Late Childhood: Longitudinal Approaches, Kalverboer A, Hopkins B, Gueze R, eds. Cambridge, England: Cambridge University Press. Nichols P (1987). Minimal brain dysfunction and soft signs: the Collaborative Perinatal Project. In: Soft Neurological Signs, Tupper D, ed. Orlando, FL: Grune & Stratton, pp 179–200. Pine D, Shaffer D, Schonfeld IS (1993). Persistent emotional disorder in children with neurological soft signs. J Am Acad Child Adolesc psychiatry 32:1229–1236. Pine DS, Wasserman GA, Fried JE, Parides M, Shaffer D (1997). Neurological soft signs: one-year stability and relationship to psychiatric symptoms in boys. J Am Acad Child Adolesc Psychiatry 36:1579–1586. Reeves J, Werry J (1987). Soft signs in hyperactivity. In: Soft Neurological Signs, Tupper D, ed. Orlando, FL: Grune & Stratton. Rie E (1987). Soft signs in learning disability. In: Soft Neurological Signs, Tupper D, ed. Orlando, FL: Grune & Stratton. Schoemaker M, kalverboer A (1990). Treatment of clumsy children. In: Developmental Biopsychology: Experimental and Observational Studies in Children at Risk, Kalverboer A, eds. Ann Arbor: University of Michigan Press, pp 241–256. Shafer S, Stokman C, Shaffer D, Ng S, Schonfeld I (1986). Ten-year consistency in neurological test performance of children with focal neurological deficit. Dev Med Child Neurol 28:417–427. Statistics Sweden (1998). Statistical Yearbook of Sweden. Stockholm: Statistics Sweden. Swedish Association of Local Authorities (1997). Conditions of Living in the Swedish Municipalities and Rural Districts. Report on Facts, May 1997. Stockholm: Swedish Association of Local Authorities. Szatmari P, Brenner R, Nagy J (1989a). Asperger’s syndrome: a review of clinical features. Can J Psychiatry 34:554–560.
Annual progress in child psychiatry and child development 2000–2001
304
Szatmari P, Offord DR, Boyle MH (1989b). Correlates, associated impairments and patterns of service utilization of children with attention deficit disorder: findings from the Ontario Child Health Study. J Child Psychol Psychiatry 30:205–217. Tupper D (1987). The issues with “soft signs.” In: Soft Neurological Signs, Tupper D, ed. Orlando, FL: Grune & Stratton. van Dellen T, Vaessen W, Schoemaker M (1990). Clumsiness: Definition and Selection of Subjects. Ann Arbor, MI: University of Michigan Press. Whitmore K, Bax M (1990). Checking the health of school entrants. Arch Dis Child 65:320–326. Wilson P, McKenzie B (1998). Information processing deficits associated with developmental coordination disorder: a meta-analysis of research findings. J Child Psychol Psychiatry 39:829–840. Witmont K, Clark C (1996). Kinaesthetic acuity and fine motor skills in children with attention deficit hyperactivity disorder: a preliminary report. Dev Med Child Neurol 38:1091–1098. Wolff PH, Michel GF, Ovrut M, Drake C (1990). Rate and timing of motor coordination in developmental dyslexia. Dev Psychol 26:349–359. Wright HC, Sugden DA (1996). A two-step procedure for the identification of children with developmental coordination disorder in Singapore. Dev Med Child Neurol 38:1099–1105.
PART IV: OTHER CLINICAL ISSUES
16 Thirty-Three Cases of Body Dysmorphic Disorder in Children and Adolescents Ralph S.Albertini and Katharine A.Phillips
Objective: Body dysmorphic disorder (BDD), a preoccupation with a nonexistent or slight defect in appearance, usually begins during adolescence. Because there have been no studies of the clinical features of BDD in children and adolescents, the authors assessed these features in the largest series to date. Method: Thirty-three children and adolescents with DSM-IV BDD were assessed for demographic characteristics, phenomenology, associated psychopathology, and treatment history and response. Results: Bodily preoccupations most often focused on the skin (61%) and hair (55%). All subjects had associated compulsive behaviors, most often camouflaging (e.g., with clothing) in 94%, comparing with others (87%), and mirror checking (85%). Ninety-four percent reported impairment in social functioning and 85% in academic or job functioning because of BDD. Thirty-nine percent had had psychiatric hospitalizations, and 21% had made a suicide attempt. Ten (53%) of 19 subjects treated with a serotonin reuptake inhibitor had much or very much improvement in BDD symptoms; in contrast, 0 of eight trials with other psychotropic medications, 0 of one trial of cognitivebehavioral therapy, and one of 20 psychotherapy trials resulted in improvement. Twelve (36%) subjects received surgical, dermatological, or dental treatment, with a poor outcome in all cases. Conclusions: BDD can cause significant morbidity in children and adolescents. These preliminary data suggest that serotonin reuptake inhibitors may be an effective treatment in this age group. (J Am Acad Child Adolesc Psychiatry, 1999, 38(4):453–459.) Key Words: body dysmorphic disorder, dysmorphophobia, body image.
INTRODUCTION Body dysmorphic disorder (BDD), a distressing and impairing preoccupation with an imagined or slight defect in appearance, has been described for more than a century and reported around the world (Phillips, 1991). Available data suggest that BDD is relatively
Annual progress in child psychiatry and child development 2000–2001
306
common (Phillips et al., 1996; Simeon et al., 1995) and usually begins during adolescence. Despite receiving increasing clinical and research attention, this disorder remains virtually unstudied in children and adolescents. A growing body of evidence indicates that BDD in adults is characterized by painful and time-consuming obsessions about the perceived appearance defect as well as compulsive, time-consuming behaviors such as mirror checking, excessive grooming, and skin picking (Hollander et al., 1993; Phillips et al., 1993). Insight is generally poor, and a significant percentage of patients are delusional. BDD usually causes considerable morbidity, such as social, educational, and occupational impairment, being housebound, psychiatric hospitalization, suicide attempts, and completed suicide (Hollander et al., 1993; Phillips, 1991; Phillips et al., 1993). BDD usually begins during adolescence (Phillips et al., 1995a). In the largest published series of subjects with BDD (n=188), the mean age at onset was 16.0±7.2 years (range=4 to 43 years), with BDD beginning before age 18 in 70% of cases (Phillips & Diaz, 1997). Reported cases in children and adolescents suggest that the clinical features of the disorder in this age group are generally similar to those in adults (Braddock, 1982; Cotterill, 1981; El-Khatib & Dickey, 1995; Hay, 1970; Heimann, 1997; Phillips et al., 1995a; Sondheimer, 1988). As in adults, BDD in children and adolescents may lead to impaired functioning, such as poor grades (Phillips et al., 1995a), stopping sports and other activities (El-Khatib & Dickey, 1995), excessive school absences (Albertini et al., 1996), quitting high school (Phillips et al., 1995a), social withdrawal (El-Khatib & Dickey, 1995), and being housebound (Cotterill, 1981), any of which may adversely affect development. It may also result in psychiatric hospitalization, suicidal ideation, and suicide attempts (Cotterill, 1981; Phillips et al., 1995a). Several case reports suggest that serotonin reuptake inhibitors (SRIs) may be effective in decreasing BDD symptoms and improving functioning in children and adolescents (Heimann, 1997; Sondheimer, 1988). Nearly all of the literature on BDD in children and adolescents, however, consists of case reports, with the largest published series in this age group limited to four cases (Phillips et al., 1995a). In this study we systematically assess 33 consecutive children and adolescents with DSM-IV-defined BDD. We describe demographic characteristics, phenomenology, associated features, associated psychopathology, and treatment history and response. To our knowledge, this is the largest series of children and adolescents with BDD that has been described.
METHOD Subjects Thirty-three consecutively seen children and adolescents (aged 17 years or younger) who met DSM-IV criteria for BDD were included in the study. The majority of subjects were referred to a BDD clinical and research program for evaluation and/or treatment from the inpatient and outpatient services of a private psychiatric hospital, whereas others were referred by community therapists or parents or were self-referred. All subjects met DSMIV criteria for BDD, which are as follows: (a) preoccupation with an imagined defect in
Thirty-three cases of body dysmorphic
307
appearance; if a slight physical anomaly is present, the person’s concern is markedly excessive; (b) the preoccupation causes clinically significant distress or impairment in social, occupational, or other important areas of functioning; and (c) the preoccupation is not better accounted for by another mental disorder (e.g., dissatisfaction with body shape and size in anorexia nervosa). Patients with preoccupations that were delusional (delusional disorder, somatic type) were included in the study because available data suggest that delusional and nondelusional BDD are variants of the same disorder (Phillips et al., 1994), and they are double-coded in DSM-IV. All subjects and their parents/guardians signed statements of informed consent. Assessments Thirty-one subjects, those aged 12 through 17 years, were administered the Structured Clinical Interview for DSM-III-R (SCID) (Spitzer et al., 1992; Williams et al., 1992) to obtain information on demographic characteristics and associated psychopathology. Two subjects, aged 6 through 11 years, were administered the Schedule for Affective Disorders and Schizophrenia for SchoolAge Children-Present and Lifetime version (KSADS-PL) (Kaufman et al., 1997). Because neither the DSM-III-R SCID nor the KSADS-PL include BDD, this disorder was diagnosed with a brief, reliable, semistructured diagnostic module based on the SCID that diagnoses BDD according to DSM-IV criteria (Phillips et al., 1995b) and has adequate interrater reliability in adolescents (Albertini & Phillips, unpublished data, 1997). Subjects were also given a semistructured instrument, the BDD Data Form (Phillips, unpublished, 1991), which has been used in studies in adults (Phillips et al., 1993, 1994), to obtain additional demographic and clinical information on BDD such as data on course of illness, body areas of concern, associated behaviors, history of suicide attempts and hospitalization, and psychiatric and nonpsychiatric treatment history. Subjects aged 16 and younger were assessed using the Children’s Global Assessment Scale (Shaffer et al., 1983). Severity of BDD symptoms during the previous week was assessed with the YaleBrown Obsessive Compulsive Scale Modified for Body Dysmorphic Disorder (BDDYBOCS) (Phillips et al., 1997). This scale is a clinician-administered, semistructured, 12item adaptation of the Yale-Brown Obsessive Compulsive Scale (Goodman et al., 1989). The BDD-YBOCS assesses BDDrelated obsessional preoccupation (five items), compulsive behavior (five items), insight (one item), and avoidance (one item) during the preceding week. Scores range from 0 through 48. The adult version is reliable and valid (Phillips et al., 1997), and the adolescent version has excellent inter-rater reliability (Phillips and Albertini, unpublished data, 1997). Twenty-four consecutive subjects were assessed with the Brown Assessment of Beliefs Scale (Eisen et al., 1998), which was added later in the study. This is a reliable and valid seven-item, semistructured, clinicianadministered scale that assesses degree of delusionality (insight) during the previous week and also classifies subjects as delusional or nondelusional. Scores range from 0 through 24. The adolescent version has preliminary evidence of excellent interrater reliability (Phillips & Albertini, 1997). For some variables, data were not obtained for all 33 subjects because the variable was added later in the study. Information on treatment history (psychiatric and nonpsychiatric treatments, such as
Annual progress in child psychiatry and child development 2000–2001
308
surgical and dermatological treatment) was obtained from subjects and parents with the previously described BDD Data Form, and treatment response was assessed with the Clinical Global Improvement Scale (CGI) (Guy, 1976). Medication doses that we considered adequate were based on clinical experience with adults, although what constitutes an adequate medication dose for treating BDD in adults or children and adolescents has yet to be empirically determined. The daily SRI doses that we considered adequate for the purpose of this study were the following: fluvoxamine 150 mg, fluoxetine 40 mg, paroxetine 40 mg, sertraline 150 mg, and clomipramine 150 mg. We required a trial duration of at least 10 weeks, based on clinical experience as well as available research data in adults (Phillips et al., 1998). Seventeen (57%) of the 30 medication trials considered adequate were conducted by other clinicians, for which treatment response was assessed retrospectively and corroborated when possible by chart review and clinician interview. We conducted the remaining trials and assessed them prospectively with the CGI.
RESULTS Of the 33 children and adolescents, three (9%) were male and 30 (91%) were female. They had a mean age of 14.9 ± 2.2 years (range=6 to 17 years). Thirty-two (97%) were white and one (3%) was African American. Twentyfive (76%) were psychiatric outpatients, four (12%) were inpatients, and four (12%) were not in treatment at the time of evaluation. BDD symptoms consisted of excessive concerns with a wide variety of body parts. The most common body areas of concern were skin (61%, n=20); hair (55%, n=18); weight (48%, n=16); ugly face (39%, n=13); teeth (30%, n= 10); and legs and nose (27%, n=9 each). Additional areas of concern were the stomach (24%, n=8); breasts/pectoral muscles (21%, n=7); hips (18%, n= 6); body build (15%, n=5); lips, feet, and buttocks (12%, n=4 each); height, arm/wrist, and forehead (9%, n=3 each); eyes, eyebrows, knees, fingers, and hands (6%, n=2 each); and face size/shape, chin, jaw, head size/shape, neck, shoulders, cheeks, and toes (3%, n=1 each). Examples of complaints were looking “ugly,” acne, scarring, hair “not being right,” “big lips,” “short legs,” “gaps” between teeth, or asymmetric jaw muscles. Although weight concerns were common, no subject was concerned with body weight alone. Some individuals (25%, n=8) focused on perceived asymmetry as a problem (e.g., “uneven” hair, one breast larger than the other). The mean number of body parts of concern was 4.8±2.2 (range=1 to 8). There were statistically significant correlations between hips and legs (r=0.40, p=.02) and weight and hair (r= 0.42, p=.02). All body parts of concern appeared to the interviewers to be normal or to have only minimal anomalies. All subjects reported significant distress over their perceived defect, with seven (25%) reporting moderate distress, 17 (61%) severe distress, and 3(11%) extreme and disabling distress on the BDD-YBOCS. As several subjects stated, “I’m tormented by my looks,” “My life is like hell on earth,” and “I wish the whole world was bald so I wouldn’t have to worry about my hair.” Most subjects (68%, n=19) spent more than 3 hours a day thinking about their defect; some said it was virtually all they thought about.
Thirty-three cases of body dysmorphic
309
Twelve (50%) of the 24 subjects assessed with the Brown Assessment of Beliefs Scale had beliefs that were delusional (that is, they were 100% convinced that their perception of the defect was accurate and undistorted). Twentysix (79%) subjects expressed ideas or delusions of reference solely attributable to BDD—for example, thinking that people were laughing at them because of how they looked. All subjects performed BDD-related behaviors (mean number of behaviors =4.5±2.2; range=1 to 9) (Table 16.1). The most common behavior (94%, n =30) was camouflaging—for example, covering the perceived defect with clothes, makeup, or a hat. Eighty-seven percent (n=27) frequently compared themselves with others, and 85% (n=28) excessively checked their appearance in mirrors or other reflecting surfaces (e.g., checking their teeth in the chrome on chairs or using small mirrors during school lectures). A majority (73%, n=24) repeatedly questioned others about their appearance, one patient asking, “Mom, do you think the gap between my front teeth has gotten any wider?” up to 30 times a day. Fifty-nine percent (n=19) excessively groomed, and nearly half dieted, usually in an attempt to change the size or shape of a particular body part (e.g., “sunken” cheeks or a “big stomach”) rather than to change their weight. Thirteen (39%) compulsively picked their skin in an attempt to improve its appearance. Twentyfive percent of patients performed these behaviors for more than 8 hours a day. The mean BDD-YBOCS score for the five items assessing BDD behaviors was 13.5 ± 2.6, whereas mean score for the 5 items assessing BDD preoccupations was 12.1±3.4.
TABLE 16.1 Clinical Features of 33 Children and Adolescents with BDD
BDD-related behaviors No. of body parts of concern Camouflaging Clothes Posture/body position Makeup Hand Hair Hat Comparing with others Mirror checking Reassurance seeking Grooming Dieting Skin picking Mirror avoidance Course Age at onset (yr) Range Duration of illness (yr) Continuous
4.8±2.2 30 (94) 22 (79) 20 (65) 20 (65) 15 (52) 11 (44) 4 (13) 27 (87) 28 (85) 24 (73) 19 (59) 13 (46) 13 (39) 10 (32) 11.8±2.6 5–17 3.2±2.5 32 (97)
Annual progress in child psychiatry and child development 2000–2001
310
Episodic 1 (3) Improving 0 (0) 4 (13) Steady Worsening 27 (87) Impairment Social interference 31 (94) 28 (85) Academic/job interference History of hospitalization 13 (39) History of violence because of BDD 11 (38) 22 (67) History of suicidal ideation 7 (21) History of suicide attempts a 5 (16) Housebound because of BDD Severity 30.6±6.2 BDD-YBOCS scoreb GAF score 44.9±11.9 C-GAS score 40.3±9.6 Insight 18.7±4.0 Brown Assessment of Beliefs Scale scorec Ideas/delusions of reference 26(79) Note. Values are expressed as n (%) or mean±SD. Includes subjects with delusional BDD. BDD =body dysmorphic disorder; BDD-YBOCS=Yale-Brown Obsessive Compulsive Scale Modified for BDD; C-GAS=Children’s Global Assessment Scale; GAP=Global Assessment of Functioning. aCompletely housebound for at least 1 week. bScore on the first 10 items=25.6±5.3. cMost subjects had poor or absent insight. Nearly all subjects experienced significant impairment in functioning as a result of their BDD symptoms (Table 16.1). Ninety-four percent (n=31) experienced social interference because of embarrassment and shame over their appearance. One attractive high school senior had never been to a single athletic or social event—even avoiding her own birthday celebration—because she believed that people would laugh at her “ugliness.” Many avoided making friends and dating. Eighty-five percent (n=28) reported that their appearance obsessions and related behaviors (e.g., frequently checking their hair in a compact mirror during class) interfered significantly with their academic performance. Others avoided (and failed) gym class, often because they were too embarrassed to be seen in gym clothes. Some avoided working after school or attending school. In addition to the 39% (n=13) who temporarily missed school because of hospitalization, an additional 18% (n=6) dropped out of school (3% [n=1] dropped out temporarily and 15% [n=5] dropped out permanently) because of BDD symptoms. One middle school student had missed more than 100 days of the previous school year because of BDD. As shown in Table 1, subjects had a notably high rate of suicidal ideation and suicide attempts, as well as physical violence because of BDD symptoms
Thirty-three cases of body dysmorphic
311
(e.g., hitting themselves or destroying furniture out of frustration because “I can’t get my hair right”). As shown in Table 16.2, the most common comorbid disorder was major depression, followed by obsessive-compulsive disorder (OCD) and social phobia. In most cases, the onset of social phobia preceded that of BDD by at least 1 year (80%, n=8), whereas OCD preceded onset of BDD by at least 1 year in 40% (n=4) of cases, and major depression preceded onset of BDD by at least 1 year in only 17% (n=4) of cases. Most subjects (82%, n=27) had received psychiatric treatment. Twentytwo (67%) had received pharmacotherapy (a total of 30 trials that we considered probably adequate), 18 (55%), had received individual psychotherapy, two (6%) had received group therapy, and one (3%) had received cognitivebehavioral therapy. However, subjects did not receive treatment for an average of more than 2 years (2.7 ± 3.0 years) after BDD onset, which was 11.8 ± 2.6 years. Even then, many did not divulge their BDD to their treater. Ten (53%) of 19 subjects treated with an SRI had much or very much improvement in BDD symptoms, and 10 (45%) of 22 SRI trials led to much or very much improvement in BDD symptoms (Table 16.3). Of the 13 SRI trials conducted by the authors, 8 (62%) resulted in much or very much improvement in BDD symptoms. Six (43%) of 14 SRI trials in delusional patients led to much or very much improvement. One delusional patient responded to an SRI after an unsuccessful trial with an antipsychotic. Mean time to response was 8.0 ± 3.9 weeks (range=4 to 16 weeks). Despite the relatively high SRI doses often used, the SRIs were generally very well tolerated. Only 2 patients discontinued an SRI because of side effects.
TABLE 16.2 Associated Psychopathology of 33 Children and Adolescents with BDD
Lifetime DSM-III-R Diagnosis Mood disordersa Major depression Bipolar disorder I Bipolar disorder II Dysthymia Psychotic disordersb Delusional disorder, somatic type Schizophrenia Schizoaffective disorder Psychotic disorder NOS Anxiety disordersa Panic disorder Agoraphobia Social phobia Simple phobia Obsessive-compulsive disorder
Current Lifetime n (%) n (%) 28 (85) 30 (91) 23 (70) 24 (73) 3 (9) 3 (9) 2 (6) 3 (9) 1 (3) 1 (3) 12 (36) 12 (36) 5 (15) 5 (15) 0 (0) 0 (0) 0 (0) 0 (0) 7 (21) 7 (21) 20 (61) 21 (64) 3 (9) 5 (15) 0 (0) 0 (0) 10 (30) 10 (30) 4 (12) 5 (15) 12 (36) 13 (39)
Annual progress in child psychiatry and child development 2000–2001
312
Substance abuse/dependence 1 (3) 2 (6) Alcohol 1 (3) 1 (3) 0 (0) 1 (3) Other drug a 1 (3) 2 (6) Eating disorders Anorexia nervosa 0 (0) 2 (6) Bulimia nervosa 1 (3) 1 (3) Note. Includes the delusional variant of BDD. BDD=body dysmorphic disorder; NOS=not otherwise specified. aThe total is less than the sum of the individual disorders because some subjects had more than one disorder in a given category. bPsychotic symptoms were in all cases entirely attributable to BDD. TABLE 16.3 Open Medication Treatment Data in 33 Children and Adolescents With BDD
Fluoxetine Paroxetine (n=12) (n=4)
Settraline CMI Fluvoxamine Non(n=4) (n=1) (n=1) SRIs (n=8)
SRIs Much or very 58 (7) 25 (1) 25 (1) 100 0 0 much improved 2.4±1.0 2.8±0.5 3.0±1.4 1.0±0 3.0±0 4.0±0 CGI change for BDDb Dose (mg/day) 52.5±19.6 50.0±11.5 181.3±23.9 175 250 — 20–80 40–60 150–200 Range Time to 7.4±3.0 8.0 16.0 4.0 NA NA response (wk) 4–12 Range Note. These 33 children/adolescents received a total of 22 SRI trials and eight trials with other medications. We treated 13 of these 33 children/ adolescents outselves and conducted 13 of the 22 SRI trials. Values are expressed as % (n), mean±SD, and range. BDD=body dysmorphic disorder; CGI=Clinical Global Improvement Scale; CMI=clomipramine; NA=not available. an’s refer to number of medication trials. b1=very much improved; 2=much improved; 3=minimally improved; 4=no change on BDD CGI. SRI responders usually experienced a decrease in preoccupation, distress, and compulsive behaviors, as well as improved functioning. Many were able to resume normal activities with peers and improve school attendance and performance. Several who had dropped out of school because of BDD symptoms returned to school. Subjects
Thirty-three cases of body dysmorphic
313
who responded to an SRI usually had a sustained response for periods up to several years. Two of 3 responders who discontinued the SRI relapsed. Patients without current major depression were as likely to have response of BDD symptoms to an SRI as those with current major depression (n=3 of 5 [60%] and n=7 of 14 [50%], respectively). Patients without current comorbid OCD were also as likely to have response of BDD symptoms as those with current OCD (n=6 of 13 [46%] and n=4 of 6 [67%], respectively). No non-SRI medication was effective in decreasing BDD symptoms, although comorbid conditions improved in 2 instances. (ADHD improved with clonidine and stimulants in one patient, and depression and panic disorder improved with imipramine in another.) Although 18 (55%) subjects received other psychiatric treatments, they were generally ineffective. Only one of 20 psychotherapy (insight-oriented or supportive) treatments, 0 of one cognitive-behavioral therapies, and 0 of two group therapies resulted in significant improvement in BDD. Forty-five percent (n= 15) of the subjects sought surgical or medical treatment for their BDD, and 36% (n=12) received such treatment. Of nine dermatological treatments, two dental treatments, and one surgical procedure received, none improved BDD symptoms.
DISCUSSION These results indicate that BDD in children and adolescents is characterized by painful and time-consuming preoccupations and compulsive behaviors that are associated with significant distress and impairment in functioning. Social impairment is nearly universal and often consists of extreme self-consciousness, embarrassment, and avoidance of social interactions. A majority experience academic difficulties, and some drop out of school. A notably high percentage have suicidal ideation or attempt suicide. Our treatment data, although preliminary and uncontrolled, suggest that SRIs are often effective in decreasing BDD symptoms in children and adolescents. In contrast, nonpsychiatric treatment such as surgery and dermatological treatment, which was obtained by a substantial number of patients, was in all cases ineffective. Although we did not formally compare this series with a series of adults with BDD, it is our impression that the clinical features of BDD are generally similar in these age groups. Similarities include most body areas of concern, associated behaviors, comorbidity, and degree of preoccupation, distress, and impairment in functioning. The relatively high percentage of children and adolescents who were delusional, and the high rate of associated ideas and delusions of reference is also similar to that reported for adults. Treatment data in adults suggest that SRIs are often effective for BDD (Hollander et al., 1989; Perugi et al., 1996; Phillips et al., 1998), similar to findings in this study. One apparent difference, however, is the relatively high percentage of children and adolescents concerned with their weight. Another apparent difference is that nearly all of the subjects in this study were female, in contrast to studies in adults. Although some adult series contain more women than men (Rosen & Reiter, 1996; Veale et al., 1996), others contain more men (Hollander et al., 1993) or an approximately equal number of
Annual progress in child psychiatry and child development 2000–2001
314
women and men (Phillips & Diaz, 1997). In the Phillips and Diaz series, which is the largest published adult series (n=188), the equal sex ratio and similar age at onset of BDD in males and females suggests that the very high percentage of females in our adolescent series may reflect more treatment-seeking among adolescent females, rather than more frequent occurrence of BDD in adolescent females. The reason for the preponderance of white subjects is unclear, but findings are similar in adults (Phillips et al., 1993); it is unclear whether this finding is a characteristic of BDD or reflects referral bias owing to the fact that patients in this series were evaluated at a private psychiatric hospital. Our finding that rates of comorbid major depression, OCD, and social phobia were relatively high and similar to rates in adults with BDD raises the question of whether BDD might be related to, or a form of, these other disorders. Indeed, BDD is widely conceptualized as an OCD spectrum disorder (Hollander & Phillips, 1992; Phillips et al., 1995c). Available data suggest, however, that although BDD and OCD have many similarities, they also have some differences, including for BDD a higher rate of comorbid major depression and social phobia, poorer insight, and a higher rate of suicide attempts owing to the disorder (Phillips et al., 1998). BDD’s prominent obsessions and compulsions, as well as preliminary data indicating that SRIs but not other antidepressants may be effective for BDD suggest that BDD is not simply a form of depression. Out finding in this study that onset of BDD usually preceded that of major depression is consistent with this view. It is possible, nonetheless, that BDD is related to depression and OCD as well as social phobia (Phillips et al., 1995c). Because our treatment data are uncontrolled and were obtained by selfreport, they should be considered preliminary. However, the apparent response to SRIs in a sizable percentage of subjects is notable and consistent with results from studies in adults. Nonetheless, a sizable proportion of the subjects did not respond to SRIs, and many responders had only a partial response, also similar to findings in adults (Phillips et al., 1998). That SRI treatment was not significantly helful to more subjects may perhaps reflect their severity of illness and underscores the need for more treatment research. It is worth noting that delusional patients (who would receive a DSM-IV diagnosis of delusional disorder as well as BDD) were as likely to respond to an SRI as nondelusional patients, consistent with reports in adults (Phillips et al., 1998) and adolescents (ElKhatib & Dickey, 1995; Sondheimer, 1988). It is also interesting that one patient with delusional BDD responded to an SRI after failing to respond to an antipsychotic. Whereas relatively high SRI doses were used in this study, it remains to be empirically determined whether higher SRI doses are required for BDD than for depression and other disorders. The lengthy time to response, which is consistent with data in adults (Phillips et al., 1998), suggests that relatively long SRI trials (at least 3 months) should be conducted before concluding that the medication is ineffective. Although other treatments were generally ineffective, it is our clinical impression that although psychotherapy alone tends to be ineffective for HDD, it may be useful when combined with medication (for example, when treating the social skills deficits that can result from BDD symptoms). This impression, however, requires empirical confirmation. Although preliminary data in adults suggest that cognitive-behavioral therapy may be effective for BDD (Veale et al., 1996), the efficacy of this treatment, too, remains to be studied in children and
Thirty-three cases of body dysmorphic
315
adolescents. This study’s conclusions are affected by several methodological limitations. First, patients were recruited from a psychiatric population, which may have led to elevated rates of associated psychopathology and may have selected for a particularly distressed and impaired series of subjects. A community sample might be expected to have not only lower comorbidity rates and less functional impairment, but also a more equal sex ratio and less past treatment. In addition, interviews were done by unblinded investigators and without a control group, and the treatment data were also uncontrolled. Larger studies of children and adolescents with BDD, including treatment studies, with control groups are needed. Clinical Implications Although appearance concerns are common during adolescence and may contribute to the timing of BDD’s onset, BDD does not simply consist of normal appearance concerns, as documented in this study by the degree of preoccupation, distress, and impairment in functioning, as well as the high rate of suicidal ideation and suicide attempts experienced by our subjects. To diagnose BDD in children and adolescents, it is often necessary to inquire specifically about BDD symptoms (asking, for example, “Is there some aspect/part of your appearance that you’re really unhappy about?”), as they typically are not divulged because of embarrassment and shame. Adolescents may be particularly reluctant to divulge BDD symptoms because of self-consciousness and difficulties with confiding in adults. In conclusion, BDD in children and adolescents appears to be associated with significant morbidity, which could adversely affect psychosocial development. Although surgical, dermatological, and dental procedures appear ineffective, SRIs appear to often be effective, even if the symptoms are delusional. Although much remains to be learned about BDD in children and adolescents, it is important that it be recognized and treated in this age group, who may be particularly vulnerable to the development of this distressing, often secret, and underrecognized disorder.
ACKNOWLEDGMENT This work was supported in part by an unrestricted educational grant from Solvay Pharmaceuticals. We thank Lynne M.DeMarco, M.S.P.H., for her assistance with data analysis.
REFERENCES Albertini R, Phillips KA, Guvremont D (1996). Body dysmorphic disorder in a young child (letter). J Am Acad Child Adolesc Psychiatry 35:1425–1426. Braddock LE (1982). Dysmorphophobia in adolescence: a case report. Br J Psychiatry 140:199–201.
Annual progress in child psychiatry and child development 2000–2001
316
Cotterill JA (1981). Dermatological non-disease: a common and potentially fatal disturbance of cutaneous body image. Br J Dermatol 104:611–619. Eisen JL, Phillips KA, Baer L, Beer DA, Atala KD, Rasmussen SA (1998). The Brown Assessment of Beliefs Scale: reliability and validity. Am J Psychiatry 155:102–108. El-Khatib HE, Dickey, TO (1995). Sertraline for body dysmorphic disorder (letter). J Am Acad Child Adolesc Psychiatry 34:1404–1405. Goodman WK, Price LH, Rasmussen SA, et al. (1989). The Yale-Brown Obsessive Compulsive Scale, I: development, use, and reliability . Arch Gen Psychiatry 46:1006– 1011. Guy W (1976). ECDEU Assessment Manual for Psychopharmacology. Washington, DC: US Department of Health, Education and Welfare. Hay GG (1970). Dysmorphophobia. Br J Psychiatry 116:399–406. Heimann SW (1997). SSRI for body dysmorphic disorder (letter). J Am Acad Child Adolesc Psychiatry 36:868. Hollander E, Cohen L, Simeon D (1993). Body dysmorphic disorder. Psychiatr Ann 23:359–364. Hollander E, Liebowitz MR, Winchel R, Klumer K, Klein DF (1989). Treatment of bodydysmorphic disorder with serotonin reuptake blockers. Am J Psychiatry 146:768–770. Hollander E, Phillips KA (1992). Body image and experience disorders: Body dysmorphic and depersonalization disorders. In: Obsessive Compulsive-Related Disorders, Hollander E, ed. Washington, DC: American Psychiatric Press. Kaufman J.Birmaher B.Brent D et al. (1997). The Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime version (KSADS-PL): initial reliability and validity data. J Am Acad Child Adolesc Psychiatry 36:980–988. Perugi G, Giannotti D, Di Vaio S, Frare F, Saettoni M, Cassano GB (1996). Fluvoxamine in the treatment of body dysmorphic disorder (dysmorphophobia). Int Clin Psychopharmacol 11:247–254. Phillips KA (1991). Body dysmorphic disorder: the distress of imagined ugliness. Am J Psychiatry 148:1138–1149. Phillips KA, Atala KD, Albertini RS (1995a). Case study: body dysmorphic disorder in adolescents. J Am Acad Child Adolesc Psychiatry 34:1216–1220. Phillips KA, Atala KD, Pope HG (1995b). Diagnostic instruments for body dysmorphic disorder. In: New Research Program and Abstracts. American Psychiatric Association 148th Annual Meeting, Miami, FL. Phillips KA, Diaz S (1997). Gender differences in body dysmorphic disorder. J Nerv Ment Dis 185:570–577. Phillips KA, Dwight MM, McElroy SL (1998). Efficacy and safety of fluvoxamine in body dysmorphic disorder. J Clin Psychiatry 59:165–171. Phillips KA, Gunderson CG, Mallya G, McElroy SL, Carter W (1998). A comparison study of body dysmorphic disorder and obsessive compulsive disorder. J Clin Psychiatry 59:568–575. Phillips KA, Hollander E, Rasmussen SA, Aronowitz BR, DeCaria C, Goodman WK (1997). A severity rating scale for body dysmorphic disorder: development, reliability, and validity of a modified version of the Yale-Brown Obsessive Compulsive Scale. Psychopharmacol Bull 33:17–22.
Thirty-three cases of body dysmorphic
317
Phillips KA, McElroy SL, Hudson JI, Pope HG Jr (1995c). Body dysmorphic disorder: an obsessive-compulsive spectrum disorder, a form of affective spectrum disorder, or both? J Clin Psychiatry 56(suppl 4):41–52. Phillips KA, McElroy SL, Keck PE Jr, Hudson JI, Pope HG Jr (1993). Body dysmorphic disorder: 30 cases of imagined ugliness. Am J Psychiatry 150:302–308. Phillips KA, McElroy SL, Keck PE Jr, Hudson JI, Pope HG Jr (1994). A comparison of delusional and nondelusional body dysmorphic disorder in 100 cases. Psychopharmacol Bull 30:179–186. Phillips KA, Nietenberg AA, Brendel G, Fava M (1996). Prevalence and clinical features of body dysmorphic disorder in atypical major depression. J Nerv Ment Dis 184:125– 129. Rosen JC, Reiter J (1996). Development of the body dysmorphic disorder examination. Behav Res Ther 34:755–766. Shaffer D, Gould MS, Brasic J, et al. (1983). A children’s global assessment scale. Arch Gen Psychiatry 40:1228–1231. Simeon D, Hollander E, Stein DJ, Cohen L, Aronowitz B (1995). Body dysmorphic disorder in the DSM-IV field trial for obsessive-compulsive disorder. Am J Psychiatry 152:1207–1209. Sondheimer A (1988). Clomipramine treatment of delusional disorder, somatic type. J Am Acad Child Adolesc Psychiatry 27:188–192. Spitzer RL, Williams JB, Gibbon M, First MB (1992). The Structured Clinical Interview for DSM-III-R (SCID), I: history, rationale, and description. Arch Gen Psychiatry 49:624–629. Veale D, Gournay K, Dryden W, et al. (1996). Body dysmorphic disorder: a cognitive behavioural model and pilot randomized controlled trial. Behav Res Ther 34:717–729. Williams JBW, Gibbon M, First MB, et al. (1992). The Structured Clinical Interview for DSM-III-R (SCID), II: multisite test-retest reliability, Arch Gen Psychiatry 49:630– 636.
PART IV: OTHER CLINICAL ISSUES
17 Case Series: Catatonic Syndrome in Young People David Cohen, Martine Flament, Pierre-Francois Dubos, and Michel Basquin
This article reviews all recent (1977–1997) reports on catatonic adolescents and summarizes the nine consecutive cases seen at the authors’ institution during the past 6 years. Catatonia occurs infrequently in adolescents (0.6% of the inpatient population), but it appears to be a severe syndrome in adolescents of both sexes. Diagnoses associated with catatonia are diverse, including in this series: schizophrenia (n=6), psychotic depression (n=1), mania (n=1), and schizophreniform disorder (n=1). Two patients had a previous history of pervasive developmental disorder. In the literature, catatonia was also reported in children with organic condition (e.g., epilepsy, encephalitis). Therapeutic management depends on the specific causes, but several points need to be stressed: (a) the frequency of neuroleptic-induced adverse effects; (b) the potential efficacy of sedative drugs on motor signs; (c) the possible use of electroconvulsive therapy; and (d) the necessity to manage family reactions and fears, which are frequent causes of noncooperation. It is concluded that catatonia is an infrequent but severe condition in young people. Although symptomatology, etiologies, complications, and treatment are similar to those reported in the adult literature, findings differ with regard to the female-male ratio and the relative frequencies of associated mental disorders. (J Am Acad Child Adolesc Psychiatry, 1999, 38(8):1040–1046.) Key Words: catatonia, adolescence, electroconvulsive therapy, neuroleptics, benzodiazepines.
INTRODUCTION Catatonia has become rare with the era of psychotropic drugs. However, scanty reports indicate that the syndrome can occur in adolescence and may be life threatening (Ainsworth, 1987; Kish et al., 1990). With the DSM-IV, clinicians should be able to recognize the syndrome outside the limits of a diagnosis of schizophrenia (American Psychiatric Association, 1994). Contrary to previous versions of DSM, in which catatonia
Case series
319
was associated only with schizophrenia, in the DSM-IV catatonia remains associated with schizophrenia (F20.2× [295.20]) but also appears in a separate class as “catatonic disorder due to a general medical condition” (F06.1 [293.89]) and as a specifier of mood disorders “with catatonic features.” In the ICD-10 (Organisation Mondiale de la Sante, 1994), catatonia is associated only with schizophrenia, and stupor with melancholia. We sought to investigate the incidence, clinical features, and course of catatonia in young people by: (a) systematically searching all published literature for cases of catatonia reported in adolescents during the past 20 years; and (b) describing all consecutive cases of catatonia admitted to our child and adolescent psychiatry department during the past 6 years.
CATATONIA IN YOUNG PEOPLE: A REVIEW OF THE LITERATURE We searched the English and French literature between 1977 and 1997 for all cases of catatonia reported in adolescents. Published reports are summarized in Table 17.1. The most striking findings from this 20-year review are: (a) the very limited number of cases reported (17 girls and 25 boys); (b) the diversity of diagnoses associated with catatonia; and (c) the lack of systematic studies. In accordance with the adult literature (Rosebush et al., 1990), mood disorders seem to be the most frequent diagnosis associated with catatonia, if we consider the number of cases reported. However, although a majority of catatonic adults are women, the converse seems to be true for adolescents. Electroconvulsive therapy (ECT) seems to remain the preferred treatment for catatonic depression. The use of lithium has been reported in only one case, although catatonic depressed adolescents frequently exhibit a bipolar course (Table 17.1). In the case of catatonic schizophrenia, antipsychotic drugs have been recommended (Fink, 1994) but clinical inefficacy is frequent, and neuroleptics have been associated with a risk of malignant catatonia (Philbrick & Rummans, 1994). ECT usually is a secondary option in treating catatonic schizophrenia, yet its efficacy is sometimes impressive (Table 17.1).
TABLE 17.1 Literature Reports (1977–1997) of Catatonia in Young People
Authors (Year)
Kramer (1977)
SexAge
Main Clinical Diagnosis Features Organic Condition (n=8) Epilepsy F 13 Menstrual psychosis F 10 Acute onset Epilepsy
Shah &Kaplan (1980) Ainsworth (1987) F 14 Malignant Encephalitis? catatonia Davis & Borde M 12 Progressive onset Wilson disease (1993) Maxwell et al. F 17 Acute onset, Ecstasy
Efficient Treatment Phenytoin Phenytoin (Patient died) Cu2 chelator, diazepam Conservative
Annual progress in child psychiatry and child development 2000–2001
(1993) Primavera et al. (1994) Rosebush et al. (1995)
hyponatremia F 17 Acute onset, hyponatremia F 17 Acute onset, schizophrenia M 17 Progressive onset
Pheterson et al. M (1985) Black et al. (1985) M Barnes et al. (1986) F a
F F Powell et al. (1988) M Nolen and Zwaan M (1990) Schneekloth et al. M (1993) M Cizadlo & F Wheaton (1995) Moise & Petrides M (1996) M F M F M Yeung et al. (1996) M Cohen et al. (1997b)
F F
Schizophrenia
intoxication Ecstasy intoxication Epilepsy
Tay-Sachs disease Mood Disorder (n=19) 16 Severe catatonia Psychotic depression 11 Stupor Depression 13 Recurrent Psychotic catatonia depression 17 Recurrent Psychotic catatonia depression 11 Familial catatonia Idiopathic catatonia 13 Stupor Bipolardepressed 18 Acute malignant Psychotic catatonia depressed NR Posturing, staring, Bipolar-? mutism NR Rigidity, repetitiveBipolar-? behavior 8½ Severe catatonia Psychotic depression 16 Staring, posturing Psychotic depression 16 Mutism, staring, Catatonia posturing 18 NMS Bipolar-? 16 Mutism, staring Catatonia 17 Staring, posturing, Bipolar-? mutism Bipolar-? 16 Mutism, waxy flexibility… 17 Stupor Psychotic depression 15 Malignant Bipolarcatatonia depressed 16 Stupor, psychotic Bipolardepression depressed
320
Conservative Phenobarbitone Lorazepam Lithium ECT Spontaneous recovery ECT ECT ECT ECT, dantrolene ECT ECT ECT ECT ECT Conservative ECT ECT ECT ECT ECT ECT
Case series
321
(n=6) Guttmacher and M 14 NR Schizophrenia ECT Cretella (1988) Marneros and Jäger M 14 Stupor, acute Schizophrenia Lorazepam,b (1993) onset Haloperidol Cohen et al. M 16 Mutism, waxy Schizophrenia ECT (1997b) flexibility… Schizophrenia ECT Walter and Rey M NR NR (1997) Schizophrenia ECT M NR NR Schizophrenia ECT M NR NR History of Pervasive Developmental Disorder (n=5) ECT Revuelta et al. F 18 NMS Depressionc (1994) Reamulto & Nortriptyline, M 21 Stupor Depression August (1991) haloperidol M 20 Stupor BipolarNortriptyline depressed Schizophrenia Haloperidol M 16 Posturing, catalepsy… NR Zaw & Bates M 14 Mutism, Zolpidemd posturing… (1997) Other (n=4) a F 16 NMS, depression Acute psychosis (Patient died) Kish et al. (1990) F 14 Mutism, rigidity…Atypical Rosebush et al. Lorazepamb a psychosis (1990) Ungvari et al. M 18 Stuporous reaction Conversion Lorazepam (1994) catatonia Wolanczyk et al. M 14 Stupor, mutism, Anorexia, Perphenazine (1997) incontinence depression Note. ECT=electroconvulsive therapy; NMS=neuroleptic malignant syndrome; NR=not reported. aAdolescents included in adult series. bEffective only on psychomotor symptoms. cDiagnosis not specified by authors but inferred on clinical description. dUsed as a diagnostic test. In adults, many authors have recently advocated the use of sedative drugs, barbiturates (McCall et al., 1992), benzodiazepines (Fricchione et al., 1983), or zolpidem (Thomas et al., 1997). Although catatonic signs usually improve, these drugs rarely modify the associated mental disorder (Fink, 1994). In the adolescent reports, only nine subjects received a sedative drug: two depressed patients (Nolen & Zwaan, 1990; Yeung et al., 1996) and one anorectic patient (Wolanczyk et al., 1997) did not respond, whereas six
Annual progress in child psychiatry and child development 2000–2001
322
other patients improved (Table 17.1). Given the limited adverse effects of sedative drugs, they should be considered as a first-line treatment for catatonic symptoms.
CASE REPORTS During the period 1991 to 1997 in the Department of Child and Adolescent Psychopathology at La Pitie-Salpétriêre Hospital in Paris, every adolescent was systematically assessed for catatonic signs. The diagnosis of catatonia was made in the presence of at least one motor sign (catalepsy, stupor, posturing, waxy flexibility, staring, negativism, stereotypies, psychomotor excitement, automatic compulsive movements, or muscular rigidity) in combination with another motor sign or with a symptom in the following list: social withdrawal, mutism, mannerism, echopraxia, echolalia, incontinence, verbigeration (meaningless and stereotyped repetition of words), schizophasia (scrambled speech), acrocyanosis (cyanosis of the extremities), or refusal to eat. Our list of symptoms is similar to the Bush-Francis catatonia rating scale (Bush et al., 1996), except that we added refusal to eat, incontinence, acrocyanosis, and schizophasia, according to Ey’s earlier description (1950). Each patient was examined by a pediatrician and a child psychiatrist, and every effort was made to obtain a psychiatric or a medical diagnosis. Investigations always included a semistructured family interview to assess personal and family history of both medical and psychiatric disorders, routine hematological tests, electroencephalography, and neuroimaging. Depending on both psychiatric and medical examinations, or refractoriness to pharmacotherapy, cerebrospinal fluid examination and appropriate screening tests for the known metabolic causes of the syndrome could be performed. A total of nine cases of catatonia in adolescents were seen during the designated time period. This constituted 0.6% of all hospitalized adolescents. Table 17.2 summarizes the main clinical characteristics and the therapeutic management of the nine consecutive cases. Case 1 and 3 are described in more detail. Those two cases illustrate how difficult it may be to distinguish severe obsessive-compulsive disorder from the early stage of catatonic schizophrenia, when the patient is not able to express his or her actual intrapsychic experience. Case 1 C. was a 13-year-old boy admitted to the unit because of paralytic catatonia. C. was the youngest of 3 children. His 22-year-old sister, 18-year-old brother, and parents were healthy. Family history was remarkable for many deaths at a young age among his mother’s relatives and severe psychiatric conditions among his father’s relatives (his grandmother committed suicide and one uncle had schizophrenia). C. had been treated for Asperger’s syndrome since age 4 years. He had followed the regular curriculum of a supportive private school and attended psychotherapy twice a week until age 11 years. A few months before
Case series
323
TABLE 17.2 Clinical Characteristics and Therapeutic Management of All Catatonic Adolescents Consecutively Hospitalized During the Period 1991–1997 at La Salpétriêre Hospital
CaseSexAgePrevious History 1 M 13 PDD
2
3
4
Catatonic DiagnosisComplicationsInefficient E T Symptoms Treatment Catalepsy, SCZ Skin injury Clomipramine, A rigidity, lesions, NLP- chlorpromazine,p waxy induced extra haloperidol, th flexibility, pyramidal thioridazine, negativism, state lorazepam enuresis, stereotypies, automatic movements, withdrawal, refusal to eat, echopraxia, echolalia, acrocyanosis, posturing M 13 HeterozygousStupor, waxy SCZ Malnutrition H sickle cell flexibility, disease, CDD negativism, excitement, incontinence, mutism, withdrawal, refusal to eat M 16 PDD-NOS Negativism, SCZ NLP-induced Fluoxetine, T excitement, extra chlorpromazine, stereotypies, pyramidal haloperidol automatic state movements, refusal to eat M 16 SCZ Schizophasia,SCZ NLP-induced Haloperidol, T refusal to eat, extra amisulpride, waxy pyramidal flupentixol, flexibility, state clorazepate rigidity, stereotypies, withdrawal, echolalia, incontinence,
Annual progress in child psychiatry and child development 2000–2001
324
posturing verbigeration 5 M17
Catalepsy, SCZ Dehydration Amisulpri posturing, cyamemaz waxy flexibility, mutism, incontinence, withdrawal, excitement, verbigeration Catalepsy, BipolardepressedDehydration,Cyamemazine,ECT 6bF 15 stupor, skin injury clomipramine, posturing, lesions, amisulpride, rigidity, malignant alprazolam waxy catatonia flexibility, starting, negativism, withdrawal, refusal to eat, mutism 7 F 14SexualExcitement, Bipolarmanic Amisulpri cyamemaz abuse mutism, incontinence, lithium food refusal Amisulpri 8 M15SCZ Posturing, SCZ stereotypies, automatic movements, verbigeration 9 F 14 Posturing, SCZ form NLPClorazepate, Risperidon automatic disorder induced amisulpride, lithium movements, extra loxapine rigidity, pyramidal mutism, state refusal to eat Note. CDD=developmental coordination disorder; ECT=electroconvulsive therapy NLP=neuroleptic; NOS=not otherwise specified; PDD=pervasive developmental disorder; SCZ=schizophrenia; SCZform=schizophreniform. aPack therapy=envelopment in damp sheets for 1-hour sessions with patient expressing cenesthesic sensations and spontaneous fantasies.
Case series
325
bCohen et al., 1997a.
admission, C. developed severe compulsions; he was given a diagnosis of obsessivecompulsive disorder and received clomipramine. His symptoms, however, were the beginning of his catatonic state. At the time of admission, the catatonic syndrome was extreme because of the delay before hospitalization, which was a consequence of the parents’ unusual tolerance for the symptoms. C. exhibited catalepsy with waxy flexibility and posturing, sometimes lasting several hours. He also had muscular rigidity; negativism; stereotyped movements of the mouth, hands, and shoulders; enuresis; and extreme acrocyanosis. C. was not mute but exhibited blunted affect, disorganized thoughts, neologisms, and inappropriate smiles. He disclosed prominent delusions and hallucinations, including voices of orders and commands, tactile and olfactory hallucinations, and visions of skeletons. His compulsions appeared to be related to his delusions. When the diagnosis of catatonic schizophrenia was confirmed, clomipramine was discontinued and antipsychotics were started. The catatonia did not respond to chlorpromazine (300 mg/day), nor to haloperidol (15 mg/day) and lorazepam (5 mg/day). A relay with thioridazine (200 mg/day) provoked a severe dystonic reaction that subsided after 1 week of medication discontinuation. After 4 months, ECT was proposed to C.’s parents because of complications including skin breakdown, but they refused the treatment. Amisulpride (800 mg/day) was added to pack therapy twice a week. This new therapeutic regimen led to marked symptomatic improvement. After 6 months, C. was able to leave the hospital and to enter a day-care unit. Now 16 years old, C. still exhibits residual symptoms both of the catatonic and schizophrenic spectrum. His treatment includes amisulpride (800 mg/day), biperiden (8 mg/day), psychotherapy twice a week, and milieu therapy in a day hospital; in addition, family therapy was recently started. Case 3 A. was 16 years old when admitted to our inpatient unit for compulsive water-drinking. A. was the second of three siblings. His older sister died after 2 days of life, and his younger sister was healthy. Both parents were working, although his mother had had dysthymia for several years. At age 5 years, A. began treatment with psychotherapy for nonautistic pervasive developmental disorder. Although his intellectual skills were in the low average range, he was kept in a regular school until age 14 years because of his mother’s insistence. Then he entered a special education program, where he received vocational training. A year later he was sexually abused by another boy, and his behavior changed. A. began to isolate himself and to drink a lot of water. He was first hospitalized for 2 months, with symptoms regarded as compulsive behaviors: rituals concerning eating and drinking, and ordering rituals. He explained that he needed to drink a certain number of glasses of water to alleviate his anxiety. Fluoxetine (20 mg/day) and cyamemazine (50 mg/day) were started together with behavioral therapy for water control. As he was feeling better, he left the hospital for the summer holidays. Three months later, A. was readmitted to the unit under emergency conditions because of life-threatening water intoxication. The few days before hospitalization, his parents said he was drinking up to
Annual progress in child psychiatry and child development 2000–2001
326
14 liters of water per day. At the time of admission, he showed confusion, headache, and nausea due to a cerebral edema confirmed by brain imaging. Water restriction was sufficient to improve his confusion, but A. exhibited catatonic signs, including stereotypies imitating the movement of drinking, negativism, and catatonic excitement. He had numerous hallucinations, including orders to drink and to make specific movements resembling compulsive behaviors. He also had delusions with fears of human immunodeficiency virus infection and of poisoning. A diagnosis of catatonic schizophrenia was made, and antipsychotics were started. Although chlorpromazine (250 mg/day), then haloperidol (15 mg/day) failed to improve the patient’s condition, and induced numerous extrapyramidal effects that needed an occasional prescription of tropatepine (20 mg/day), thioridazine (350 mg/day) subsequently showed notable efficacy. A. was able to attend behavioral therapy for control of his waterdrinking, and he left the inpatient unit after 8 months to enter a day-care hospital. During a 2-year followup period, no catatonic symptoms returned, but residual signs of schizophrenia such as marked social withdrawal and lack of initiative persisted. Pharmacotherapy was kept unchanged, but behavioral therapy was no longer necessary. A. is now about to work in a supervised environment. As illustrated in Table 17.2, the main findings of the study are the following: (a) patients usually exhibited multiple symptoms among those required for a diagnosis of catatonia; (b) the associated psychiatric diagnosis was schizophrenia in six (67%) of nine cases, schizophreni-form disorder in 1 (11%) of nine cases, and mood disorder in two (22%) of nine cases; (c) neurological or other somatic complications, including malignant catatonia (case 6), were prominent in all but two cases; (d) in many instances, several different psychotropic medications had to be tried before improvement could be obtained for both the motor signs and the underlying disorder. Another striking observation was that many of the parents displayed either an excessive tolerance toward their child’s symptoms (case 1) or, on the opposite, a complete denial of his or her clinical state, resulting in frequent discharges against medical advice (cases 2, 4, and 5). Of note, in five of the nine cases, both parents were non-French natives (from Central or North Africa, Caribbean Islands, and Armenia).
DISCUSSION The review of the literature, together with the study of all consecutive cases of catatonic adolescents hospitalized in our unit during the past 6 years, confirms that catatonia can occur in both children and adolescents (Tables 17.1 and 17.2). To our knowledge, the current series is the only systematic study of catatonia in this age group. Over the past 6 years, six boys and three girls aged 13 to 17 years were admitted for a catatonic syndrome. The female-male ratio contrasts with the adult studies, in which catatonic women represent 75% of the cases (Rosebush et al., 1990). Considering that our department represents one-third of all psychiatric beds for adolescents in the Paris area, the incidence rate of catatonia in young people can be estimated to be approximately 0.16 per year per million population. In the adult population, catatonic symptoms have been estimated to occur in almost 10% of patients admitted to psychiatric inpatient facilities
Case series
327
(Rosebush et al., 1990). Regarding the causes of the syndrome, the psychiatric diagnoses observed in our study were consistent with those reported previously, although we have not found any organic diagnosis in this series. As opposed to the frequency of affective disorders reported in the literature (Table 17.1), we found a majority of schizophrenic cases. We think that this result might be closer to the clinical reality of catatonia, for several reasons: (a) we have considered all consecutive cases during a long period of time; (b) many authors might have reported catatonia due to psychotic depression because this diagnosis was not recognized as a cause of catatonia in international classifications until the DSM-IV (Fink, 1994); and (c) many of the cases of catatonia in adolescents found in the literature come from reports on the efficacy of ECT for treatment of refractory depression (Table 17.1). Regardless of the possible consequences of organic conditions, complications of catatonia in young people appear to be the same as those reported in adults. In particular, malignant catatonia can occur in adolescents, most often with a diagnosis of depression and an acute onset (Table 17.1); in those life-threatening cases, ECT should be considered. In our experience, management of family reactions is essential. Regardless of the particular psychic state of each parent, family reactions might be related to fantasies and fears of death aroused by the catatonic symptoms of the child. Indeed, several months after case 6 recovered, her father was able to express his frightening confrontation to his “unresponsive and motionless” girl. Furthermore, in our series, all adolescents (except case 9) had a family history of serious psychiatric disorders, which might have influenced parental attitudes toward the emerging symptoms in the child and expectations regarding psychiatric care. The family background needs to be recognized, inasmuch as difficult and prolonged treatment in child psychiatry is possible only with cooperative parents. Finally, the finding that many of our patients were of foreign origin suggests a possible association between ethnic or cultural factors and the clinical expression of catatonia (note that 20% of all inpatients in our department are children of non-French natives). However, because our patients were all treated in France, this observation does not support the view that catatonia is more frequent in developing countries only because of the nonavailability of antipsychotic drugs (Blumer, 1997). Clinical Implications Recognition and management of catatonia are crucial because the syndrome can be fatal. Clinicians should be able to recognize the syndrome outside the limits of a diagnosis of schizophrenia, although this is the most frequent cause in adolescents. Benzodiazepines may be efficacious in the treatment of psychomotor signs, but the treatment depends on the specific causes of the syndrome; the use of ECT may be indicated in cases of a malignant subtype or refractoriness to pharmacotherapy. Management of family reactions and fears is essential to avoid noncooperation.
Annual progress in child psychiatry and child development 2000–2001
328
ACKNOWLEDGMENTS The authors thank Judith L.Rapoport, M.D., for her helpful comments on the manuscripts, and Michèle Koenig, M.D., for her help in writing it.
REFERENCES Ainsworth P (1987). A case of lethal catatonia in a 14 year-old girl. Br J Psychiatry 150:110–112. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Barnes MP, Saunders M, Walls TJ, Kirk CA (1986). The syndrome of Karl Ludwig Kahlbaum. J Neurol Neurosurg Psychiatry 49:991–996. Black DW, Wilcox JA, Stewart M (1985). The use of ECT in children: case report. J Clin Psychiatry 46:98–99. Blumer D (1997). Catatonia and the neuroleptics: psychobiologic significance of remote and recent findings. Compr Psychiatry 38:193–201. Bush G, Fink M, Petrides G, Dowling F, Francis A (1996). Catatonia, I: rating scale and standardized examination. Acta Psychiatr Scand 93:129–136. Cizadlo BC, Wheaton A (1995). Case study: ECT treatment of a young girl with catatonia. J Am Acad Child Adolesc Psychiatry 34:332–335. Cohen D, Cottias C, Basquin M (1997a). Cotard’s syndrome in a 15 year-old girl. Acta Psychiatr Scand 95:164–165. Cohen D, Paillère-Martinot ML, Basquin M (1997b). Use of electroconvulsive therapy in adolescents. Convulsive Ther 13:25–31. Davis EJB, Borde M (1993). Wilson’s disease and catatonia. Br J Psychiatry 162:256– 259. Ey H (1950). Etudes psychiatriques. Paris: Désclée de Brouwer & Cie. Fink M (1994). Catatonia in DSM-IV Biol Psychiatry 36:431–433. Fricchione GL, Cassem NH, Hooberman D, Hobson D (1983). Intravenous lorazepam in neuroleptic induced catatonia. J Clin Psychopharmacol 3:338–342. Guttmacher LB, Cretella H (1988). Electroconvulsive therapy in one child and three adolescents. J Clin Psychiatry 49:771–788. Kish SJ, Kleinert R, Minauf M et al. (1990). Brain neurotransmitter changes in three patients who had a fatal hyperthermia syndrome. Am J Psychiatry 147:1358–1363. Kramer MS (1977). Menstrual epileptroid psychosis in an adolescent girl. Am J Dis Child 131:316–317. Marneros A, Jäger A (1993). Treatment of catatonic stupor with oral lorazepam in a 14year-old psychotic boy. Pharmacopsychiatry 26:259–260. Maxwell DL, Polkey MI, Henry JA (1993). Hyponatraemia and catatonic stupor after taking “ectasy.” BMJ 307:1399. McCall WV, Shelp FE, McDonald WM (1992). Controlled investigation of the
Case series
329
amobarbital interview for catatonic mutism. Am J Psychiatry 149:202–206. Moise FN, Petrides G (1996). Case study: electroconvulsive therapy in adolescents. J Am Acad Child Adolesc Psychiatry 35:312–318. Nolen WA, Zwaan WA (1990). Treatment of lethal catatonia with electroconvulsive therapy and dantrolene sodium: a case report. Acta Psychiatr Scand 82:90–92. Organisation Mondiale de la Santé (1994). Classification Internationale des Maladies, 10th ed. Paris: Masson. Pheterson A, Estroff T, Sweeney D (1985). Severe prolonged catatonia with associated flushing reaction responsive to lithium carbonate. J Am Acad Child Adolesc Psychiatry 24:235–237. Philbrick KL, Rummans TA (1994). Malignant catatonia, J Neuropsychiatry Clin Neurosci 6:1–13. Powell JC, Silveria WR, Lindsay R (1988). Pre-pubertal depressive stupor: a case report. Br J Psychiatry 153:689–692. Primavera A, Fonti A, Novello P, Roccatagliata G, Cocito L (1994). Epileptic seizures in patients with acute catatonic syndrome. J Neurol Neurosurg Psychiatry 57:1419–1422. Realmuto GM, August GJ (1991). Catatonia in autistic disorder: a sign of comorbidity or variable expression. J Autism Dev Disord 21:517–528. Revuelta E, Bordet R, Piquet T, Ghawche F, Destee A, Goudemand M (1994). Catatonic aiguë et syndrome malin des neuroleptiques: un cas au cours d’une psychose infantile. Encephale 20:351–354. Rosebush PI, Hildebrand AM, Furlong BG, Mazurek MF (1990). Catatonic syndrome in a general psychiatric population: frequency, clinical presentation, and response to lorazepam. J Clin Psychiatry 51:357–362. Rosebush PI, MacQueen GM, Clarke JT, Callahan JW, Starsberg PM, Mazurek MF (1995). Late-onset Tay-Sachs disease presenting as catatonic schizophrenia: diagnostic and treatment issues. J Clin Psychiatry 56:347–353. Schneekloth MD, Rummans T, Logan KM (1993). Electroconvulsive therapy in adolescents. Convulsive Ther 9:158–166. Shah P, Kaplan S (1980). Catatonic symptoms in a child with epilepsy. Am J Psychiatry 137:378–379. Thomas P, Rascle C, Mastain B, Maron M, Vaiva G (1997). Test for catatonia with zolpidem. Lancet 349:702. Ungvari GS, Leung HCM, Lee TS (1994). Benzodiazepines and the psychopathology of catatonia. Pharmacopsychiatry 27:242–245. Walter G, Key J (1997). An epidemiological study of the use of ECT in adolescents. J Am Acad Child Adolesc Psychiatry 36:809–815. Wolanczyk T, Komender J, Brzozowska A (1997). Catatonic syndrome preceded by symptoms of anorexia nervosa in a 14-year-old boy with arachnoid cyst. Eur Child Adolesc Psychiatry 6:166–169. Yeung PP, Milstein RM, Daniels DC, Bowers BB (1996). ECT for lorazepam-refractory catatonia. Convulsive Ther 12:31–35. Zaw ZF, Bates GD (1997). Replication of zolpidem test for catatonia in an adolescent. Lancet 349:1914.
PART IV: OTHER CLINICAL ISSUES
18 Toward a Developmental Operational Definition of Autism Jane E.Gillham, Alice S.Carter, Fred R.Volkmar, and Sara S.Sparrow
Traditional approaches to diagnosing autism emphasize delays in communication and socialization. Traditional diagnostic schemes typically list symptoms (e.g., lack of eye contact), but provide little guidance on how to incorporate information about developmental level in making a diagnosis. Because standardized measures of adaptive behavior can provide information about children’s communication, socialization, and other behavior relative to their age, they may be useful tools for diagnosing autism. This study investigated the ability of the Vineland Adaptive Behavior Scales to identify children with autism. Vineland scores and measures of intellectual functioning were obtained for children with autism, PDDNOS, and other developmental disorders (DD). Discriminant function analyses indicated that the autism and combined nonautism (PDDNOS and DD) groups could be differentiated on the basis of socialization, daily living skills, and serious maladaptive behaviors. Socialization alone accounted for 48% of the variance in diagnosis. Using regression analyses derived from a large normative sample, adaptive behavior scores were predicted from chronological age (CA) and mental age (MA). Socialization scores in the autism group were substantially below the level predicted from CA or MA. An index derived from the ratio of actual to predicted socialization scores correctly classified 86% of both autism and nonautism cases. Findings suggest that comparison of obtained Vineland socialization scores to those predicted by CA or MA may be useful in clarifying the diagnosis of autism. Key Words: autism, diagnosis, Vineland socialization scores.
INTRODUCTION Since Kanner’s (1943) original description of the disorder, definitions of autism have emphasized difficulties in social interactions and communication, as well as perseverative and ritualistic behaviors (American Psychiatric Association, 1980, 1987, 1994; World Health Organization, 1993). Traditional diagnostic systems have relied heavily on
Annual progress in child psychiatry and child development 2000–2001
332
categorical approaches to defining autism. For example, the definition of “social dysfunction” has included various dichotomous criteria (e.g., the presence or absence of a “qualitative impairment in social interaction”). Despite widespread agreement about the central features of autism, categorical systems differ markedly in patterns of diagnosis (Volkmar & Cohen, 1988). Categorical definitions of autism must deal with the problems of a very broad range of syndrome expression over both age and developmental level. With few exceptions, they have failed to consider criteria within a developmental context. For example, failure to show symbolic play would reflect a deficit at a mental age (MA) of 7, but not below a MA of 2 years. Communication impairment may present as echolalia and limited spontaneous speech in a 5-year-old with autism. In contrast, an adolescent or adult may speak fluently in sentences but lack understanding of more subtle aspects of social conversation (Kanner, 1943). Bishop (1989) suggested that the lack of developmental perspective in categorical approaches is confusing to professionals and parents, and may lead to inaccurate diagnoses. Sroufe and Rutter (1984), Greenspan (1990), and others have emphasized the importance of assessing stage salient tasks in assessing autism and other developmental disorders. Standardized measures have been used to develop alternative diagnostic criteria for other developmental disorders (e.g., reading disabilities, Finucci, Isaacs, Whitehouse, & Childs, 1982; Finucci, Whitehouse, Isaacs, & Childs, 1984; Yule & Rutter, 1985). These approaches have evaluated a person’s skills in relation to those expected given his/her age or developmental level. Such approaches have several advantages, including: (a) they rely on well-normed, developmentally based assessment instruments; (b) they may be more sensitive to differences in severity than categorical approaches; and (c) they may facilitate the identification of subtypes of disorders, such as autism (Baily, Philips, & Rutter, 1996). The lack of normative developmentally based assessments of social functioning has been a limitation, until recently, in applying such an approach to the study of autistic social dysfunction. Several studies have demonstrated that the Vineland Adaptive Behavior Scales (Sparrow, Balla, & Cicchetti, 1984) can be used to document delays in social and communicative development in individuals with autism (Ando, Yoshimura, & Wakabayashi, 1980; Carter et al., 1998; Freeman, Ritvo, Yokota, Childs, & Pollard, 1988; Jacobson & Ackerman, 1990; Loveland & Kelley, 1988, 1991; Rodrigue. Morgan, & Gefken, 1991; Schatz & Hamdan-Allen, 1995; Sloan & Marcus, 1981;Volkmar et a1, 1987). In a previous study, Volkmar and colleagues evaluated the ability of the Vineland Adaptive Behavior Scales to diagnose autism (Volkmar, Carter, Sparrow, & Cicchetti, 1993). Multiple regression equations were derived from a large normative sample to predict expected socialization and communication skills on the basis of chronological or mental age, parent education, and sex of the child. These regression equations were applied to two clinically disordered groups: (a) children who met DSM-III or DSM-III-R diagnostic criteria for autism; and (b) children with other developmental disorders (including mental retardation, developmental language disorders, and specific developmental disorders). Results indicated that there was a greater discrepancy between observed and predicted socialization and communication scores in the autistic group than in the developmentally disordered group. Ratios of observed scores to predicted scores were calculated and transformed into Z scores. An MA-based Z score cut point of ≤−2.2
Toward a developmental operational definition of autism
333
versus >−2.2 in the Vineland socialization domain was a powerful predictor of diagnosis. Using this cut point, 94% of the autistic group and 92% of the developmentally disordered (DD) group were correctly classified. The addition of communication and daily living skills domain scores did not increase the number of cases correctly classified (Volkmar et al., 1993). Although deficits in both communication and socialization are characteristic of the disorder, individuals with autism tend to evidence greater impairment in socialization relative to both communication and daily living skills (cf. Carter, Gillham, Sparrow, & Volkmar, 1996). The first aim of the present study was to replicate previous findings that demonstrated delays in socialization relative to chronological age (CA) and mental age (MA) could distinguish autistic from nonautistic children (Volkmar et al., 1993). This paper expands on the earlier study in two ways. First, measures of maladaptive behaviors as well as adaptive skills were included. Deviant behaviors (e.g., poor eye contact, peculiar preoccupations, self-injurious behaviors) are common and quite salient in autism. We were interested in whether the inclusion of both adaptive and maladaptive indices of social functioning would best distinguish children with autism from children without autism and what the relative or unique contribution of each would be. Second, in contrast to an examination of children with autism and children with nonpervasive developmental disorders, a third group of children who had been diagnosed with pervasive developmental disorders (PDD) other than autism were included. Because impairment in socialization is a hallmark of all PDDs, we expected that children in the PDD group would also display socialization considerably below the level expected given their CA or MA. Specifically, we predicted that children with autism would be most impaired, children with non-PDD developmental disorders would be least impaired, and children with PDD but not autism would have intermediate levels of impairment with respect to adaptive and maladaptive skills.
METHOD Participants Our sample comprised children between the ages of 4 and 13 who were assessed over a 3-year period at a university-based clinic that specializes in diagnosis and assessment of children with PDDs, including autism. All children were referred because of concerns about developmental delays. Parents, pediatricians, and/or teachers of most of the children had also expressed concerns about the possibility of autism or an autistic spectrum disorder. Children received a comprehensive evaluation which included administration of the survey form of the Vineland, tests of communicative ability, a psychiatric examination, and a test of intellectual functioning. Children were assigned to one of three groups based on the clinical diagnoses they received; an autistic group (n=44; 36 boys, 8 girls), a group diagnosed with Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS, n=21; 18 boys, 3 girls), and a group composed of developmentally disordered children who did not have autism or another PDD (n=30; 17 boys, 13 girls). Of the children in the DD group, 22 exhibited
Annual progress in child psychiatry and child development 2000–2001
334
mild or moderate mental retardation, 6 exhibited a developmental language disorder, and 2 exhibited a combination of motor skills and language delays. The preponderance of boys in the autistic and PDD-NOS groups was expected given the increased frequency of these conditions in males (Bryson, Clark, & Smith, 1988; Burd, Fisher, & Kerbeshian, 1987; Fombonne & du Mazaubrun, 1992; Lord & Schopler, 1987; Sverd, Sheth, Fuss, & Levine, 1995). Measures Adaptive and Maladaptive Behaviors. The Vineland Adaptive Behavior Scales, Survey Form, was administered. The Vineland is a semistructured interview administered to a parent or other primary caretaker of the child. The Vineland assesses children’s day-today adaptive functioning. Scores from three domains of adaptive behavior were used in the present study: (a) communication, reflecting the child’s receptive, expressive, and written language skills; (b) daily living skills, reflecting the child’s personal self-care, domestic, and community living skills; and (c) socialization, reflecting the child’s interpersonal, play or leisure skills, and coping skills. Scores from the fourth domain of adaptive behavior—motor skills—were not used as most subjects were above the ceiling age for this scale. Consistent with previous research (Volkmar et al., 1993), we used Vineland domain raw scores rather than standard scores to avoid the possibility of floor effects in this very impaired sample. Standard scores cannot be computed below 20 on the Vineland, and many individuals in the sample had very low adaptive domain standard scores. Age-equivalent scores were not used because two contiguous raw scores may correspond to the same age-equivalent score. In addition to adaptive behavior scales, the Vineland also includes a maladaptive behaviors scale which assesses behaviors that may interfere with the development of adaptive skills. Part 1 of the maladaptive behaviors scale assesses negative behaviors that are relatively minor or common, particularly in young children (e.g., impulsivity, temper tantrums, poor eye contact). Part 2 of the maladaptive behavior scale describes more serious and more unusual maladaptive behaviors (e.g., rocking back and forth, peculiar preoccupations with objects or activities, self-injurious behavior). Although the maladaptive scales are standardly administered for children age 5 and over, they were administered for all children who participated in the study, as they are routinely included as part of the clinic assessment. Part 1 and Part 2 raw scores were used separately in analyses since, in theory, they reflect very different types of behaviors. Cognitive Functioning. For the majority of children, cognitive functioning was assessed using the K-ABC (Kaufman & Kaufman, 1983). The WISC-III (Wechsler, 1991) was used to assess five children (three in the autism group, one in the PDD-NOS group, and one in the DD group). IQ scores were converted to MA equivalents as follows. For the K-ABC, procedures outlined in the scoring manual were followed. For the WISC-III, age equivalents were computed for each subtest and then averaged. Clinical Diagnosis. Clinical diagnoses were made by clinicians who were blind to the Vineland interviews and results. To be included in the autistic or PDD-NOS group, each case had to meet either DSM-III-R or DSM-IV criteria for the disorder. Children assigned a PDD-NOS diagnosis did not meet full criteria for autism but did exhibit multiple
Toward a developmental operational definition of autism
335
symptoms consistent with this diagnosis in at least two categories (e.g., the social category plus either communication or resistance to change). Note that none of the children in the PDD-NOS group would have met DSM-IV criterion for Asperger Disorder or Childhood Disintegrative Disorder. Clinicians were highly experienced in diagnosis and assessment of autism and other pervasive developmental disorders. They participated in field trials of the diagnostic criteria for both DSM-III-R and DSMIV. Diagnostic agreement during the field trials was good, with a kappa of 0.86 (Volkmar et al., 1994). Diagnostic agreement for the clinicians providing diagnoses in the present study has been high, with kappas ranging from 0.80 to 0.95 in other research (Klin, Lang, Cicchetti, & Volkmar, in press).
RESULTS Group Differences on IQ, Mental Age, and Vineland Scores ANOVAs were conducted to determine whether there were significant differences between the three groups on IQ, MA, and Vineland scores. When these ANOVAs yielded significant results, pairwise comparisons of group means were conducted using Tukey tests. Mean age, IQ scores, and MA are presented in Table 18.1. There were no significant differences in age. There were, however, significant differences in IQ scores, F(2, 92) =11.872, p<0.01, and MA, F(2, 92)=9.48, p<0.01. Pairwise comparisons revealed that the autism group had significantly lower IQ scores than the PDDNOS (p<0.05) and DD (p<0.01) groups. The PDDNOS and DD groups did not differ significantly on IQ. The autism group also had significantly lower MA scores than the developmentally disordered group (p<0.01). The PDD-NOS group did not differ significantly in MA from the autism or DD groups.
TABLE 18.1 Mean Age, Mental Age, and IQ by Groupa
Autism (n=44)
PDD-NOS (n=21)
Developmentally Disordered (n=30)
Age M(SD) 9.161 (2.45) 8.661 (2.39) 9.251 (2.46) 4.25–12.75 4.41–12.17 4.17–13.00 Range IQ score 71.902 (18.83) M(SD) 47.891 (21.73) 63.102 (23.64) 20–103 41–132 44–112 Range Mental age 4.981.2 (1.69) 6.162 (2.46) M(SD) 4.061 (1.88) 2.25–10.25 2.92–10.00 2.58–14.56 Range aWithin each domain, means with subscript 1 differ significantly from those with
Annual progress in child psychiatry and child development 2000–2001
336
subscript 2 at p<0.05. Mean Vineland domain raw scores are presented for each group in Table 18.2. A MANOVA was conducted to examine group differences in Vineland scores for the three adaptive domains (Communication, Daily Living Skills, and Socialization) and the two maladaptive scales (Part 1 and Part 2). Results of this analysis indicated there were significant differences between groups on the Vineland, F(2, 92)=21.984, p<0.01. ANOVAS conducted for each adaptive domain revealed that there were significant group differences in all Vineland domains assessed: Communication, F(2, 92)=23.11, p<0.01. Daily Living Skills, F(2, 92)=12.22, p<0.01; and Socialization, F(2, 92)=43.21, p<0.01; as well as in both types of maladaptive behaviors, Part 1: F(2, 92)=3.11, p<0.05; Part 2: F (2, 92)=7.41, p<0.01.
TABLE 18.2 Vineland Raw Scores: Mean and Standard Deviation by Groupa
Vineland Domain Communication M(SD) MA adj. mean Daily Living M(SD) MA adj. mean Socialization M(SD) MA adj. mean Maladaptive Part 1 M(SD) MA adj. mean Maladaptive Part 2 M(SD)
Autism (n=44)
PDD-NOS (n=21)
Developmentally Disordered (n=30)
49.301 (25.60) 55.73
75.712 (15.90)
82.732 (20.23)
75.28
73.60
61.001 (27.01) 79.142 (26.67)
67.60
92.372 (27.72)
78.70
83.00
36.841 (15.24) 39.44
62.382 (11.54)
66.432 (15.73)
62.21
62.74
16.211 (4.54) 15.97
12.952 (4.33)
15.001.2 (5.78)
12.97
15.34
12.181 7.622 (4.70) 6.702 (5.34) (8.11) 12.07 7.63 6.86 MA adj. mean aWithin each domain, means with subscript 1 differ significantly from those with subscript 2 at p <ø.05. With the exception of Part 1 maladaptive behaviors, MA adjusted means showed the same pattern of differences between groups.
Toward a developmental operational definition of autism
337
Pairwise Tukey tests indicated that the autism group displayed significantly poorer communication, daily living skills, and socialization and significantly more Part 2 (or more unusual and serious) maladaptive behaviors than the developmentally disordered group (all ps<0.01). The autism and developmentally disordered groups did not differ significantly on Part 1 (or more common) maladaptive behaviors. Compared with the PDD-NOS group, the children with autism displayed significantly poorer communication (p<0.01), daily living skills (p<0.05), and socialization (p<0.01), as well as significantly more Part 1 and Part 2 maladaptive behaviors (both ps<0.05). The PDD-NOS and DD groups did not differ significantly on any of the Vineland maladaptive or adaptive domain scores. To control for group differences in MA, the Vineland analyses were repeated with MA included as a covariate. With one exception, the identical pattern of findings (including for pairwise comparisons using Tukey tests) was obtained for each Vineland adaptive domain and maladaptive scale. The only exception was in Part 1 maladaptive behaviors. The F value for the ANCOVA for Part 1 maladaptive behaviors fell to the level of nonsignificant trend, F(2, 91)=2.656, p=0.076. Similarly, follow-up Tukey tests revealed only a nonsignificant tendency for the autism group to display more Part 1 maladaptive behaviors than the PDD-NOS group (p=0.067). See adjusted means in Table 18.2. MA was a significant covariate in all of these analyses except those predicting Part 1 and Part 2 maladaptive behaviors. Discriminant Function Analyses To determine which measures of adaptive and maladaptive functioning from the Vineland best predicted clinical diagnosis, a series of stepwise discriminant function analyses were conducted. The variables on which the three groups differed were included in these analyses: MA, communication, daily living skills, socialization, and maladaptive Parts 1 and 2. In the first stepwise analysis, membership in all three groups (Autism vs. PDD-NOS vs. DD) was predicted from MA and Vineland scores. Socialization, Part 2 maladaptive behaviors, and daily living skills each accounted for unique variance in the prediction of group membership. An analysis including these variables yielded χ2 (6)=75.44 (p<0.01), and accounted for approximately 50% of the variance in group membership (Table 18.3). Overall, this analysis correctly classified 72% of cases. Ninety-one percent of the autism cases, 77% of the developmentally disordered, but only 24% of the PDD-NOS cases were correctly classified. Pairwise contrasts indicated that the autistic group could be distinguished from both the DD group, F(3, 90)=20.07, p< 0.01, and the PDD-NOS group, F(3, 90)=26.70, p<0.01. However, it was not possible to distinguish the DD and PDD-NOS groups (see Table 18.4 that presents predicted group membership). As it was not possible to discriminate between the PDD-NOS and non-PDD developmentally disordered groups, a second stepwise discriminant function analysis was conducted that combined these two groups and compared them to the children with autism. Again, socialization, Part 2 maladaptive behaviors, and daily living skills each accounted for unique variance in the prediction of group membership. An analysis including these variables yielded a χ2(3)= 72.35 (p<0.01), and accounted for nearly 48%
Annual progress in child psychiatry and child development 2000–2001
338
of the variance in group membership (see Table 18.3). Overall, 92% of children were correctly classified, with correct classification of 89% of autism cases and 94% of nonautism cases (Table 18.5). MA and Part 1 maladaptive behaviors did not account for unique variance in either discriminant function analysis.
TABLE 18.3 Stepwise Disscriminant Function Analyses: Variables Distinguishing Groups
Individual Individual Partial R2 F(df) on Final Final 2 2 F(df) Entry F(df) R on Entry Partial R Three-group analysis (autism vs. PDD-NOS vs. Developmentally Disordered) ø.484 43.21 (2, 0.358 35.83 (2, Socialization 0.484 43.21 (2, b 92)b 90)b 92) ø.094 4.75 (2, 0.081 3.98 (2, Maladaptive 0.133 7.03 (2, 92) b a Part 2 91) 90)a ø.065 3.14 0.065 3.14 (2, Daily Living 0.210 12.22 (2, Skills (2,90)a 90)a 92)b Two-group analysis (autism vs. non-autism) ø.479 85.53 (1, 0.357 50.53 (1, Socialization 0.479 85.53 (1, b 93)b 91)b 93) ø.092 9.27 (1, 0.077 7.55 (1, Maladaptive 0.130 13.93 (1, b b Part 2 92) 91)b 93) ø.042 3.96 (1, 0.042 3.96 (1, Daily Living 0.185 21.08 (1, Skills 91)a 91)a 93)b ap<0.05. bp<0.01. Variable
TABLE 18.4 Classification Resulting from Discriminant Function Analyses: Three-Group Analysis Predicted Diagnosis Austism PDD-NOS Developmental Disorder Actual diagnosis n n % n % n % Autism 44 40 91 1 2 3 7 PDDNOS 21 2 10 5 24 14 67 Developmental disorder 30 4 13 3 10 23 77
Toward a developmental operational definition of autism
339
TABLE 18.5 Classification Resulting from Discriminant Function Analyses: Autism versus Nonautism Groups
Actual diagnosis Autism Non-autism
n 44 51
n 39 3
Predicted Diagnosis Autism Nonautism % n % 89 5 11 6 48 94
Signal Detection Analyses To determine the feasibility of utilizing the discriminant power of the Vineland Adaptive Behavior Scales in clinical settings, we attempted to replicate previous signal detection analyses (Volkmar et al., 1993). In those analyses, data from children who were in the Vineland and K-ABC overlap standardization sample, a nationally representative sample, were used to derive regression equations predicting Vineland domain raw scores from chronological age (CA), sex, and parent education. Similar regression equations were derived to predict domain raw scores from MA, sex, and parent education (for a detailed discussion of these procedures, see Volkmar et al., 1993). In that study, these equations were then used in a sample of children with autism or developmental disorder to examine whether deficits in autism were: (a) more severe than predicted given the chronological or mental age; and (b) whether differences could be used to predict diagnostic group. In the present study, we used these CAand MA-based regression equations to calculate predicted Vineland socialization scores for each child as the socialization domain appears to be the most discriminating in both samples. Consistent with previous analyses, ratios of observed socialization scores over expected scores were calculated. Next, Z scores were created by first computing the difference between ratio scores and the mean ratio scores in the normative sample, and then dividing by the normative ratio standard deviation. (The means and standard deviations for the normative sample are presented in Volkmar et al., 1993.) In comparison to the nonautistic children (i.e., PDD-NOS and DD groups), the children with autism exhibited greater than expected delays in socialization skills (see Figure 18.1). In order to facilitate comparison with the study by Volkmar and colleagues, the mean ratios of observed to expected scores and corresponding Z scores for all three adaptive behavior domains are presented in Table 18.6.
Annual progress in child psychiatry and child development 2000–2001
340
Figure 18.1. Distribution (percentage of group) of Z-transformed ratios of actual to predicted socialization scores for autism (n=44) and nonautism (n=51) cases. Predictions based on MA. As socialization and the maladaptive Part 2 scores accounted for the majority of variance in classifying group membership, signal detection analyses were conducted to evaluate the ability of specific MA- and age-based socialization Z scores and Part 2 cutoffs to predict whether children received a diagnosis of autism versus nonautism (DD or PDD-NOS). These analyses were not attempted for the PDD-NOS group or the developmentally delayed group as separate entities because the ANOVAs and discriminant function analyses were not able to differentiate these two groups. For more information on signal detection procedures and their application to the study of categorical diagnostic criteria for autism see Kraemer (1988) and Siegel, Vukicevic, Elliot, & Kraemer (1989). Consistent with previous research (Volkmar et al., 1993), we calculated the sensitivity and specificity of various socialization domain cut points in tenths (e.g., −5.0, −4.9, −4.8). Then, the sensitivity and specificity were corrected for the level of criterion (autism) in the sample. This step involves correcting for the probability that a randomly selected case will be autistic. The resulting kappa values (Kse and Ksp) were then plotted in the Quality Receiver Operator Curve (QROC) to identify the optimal cut point (the cut point that maximized Kse×Ksp), after correcting for the level of criterion in the sample. Finally, a tree procedure was used to determine whether the addition of maladaptive scores significantly increased the proportion of cases correctly classified.
Toward a developmental operational definition of autism
341
TABLE 18.6 Ratios of Actual Raw Scores to those Predicted and Corresponding Z Scorea
Autism Group (n=44) Ratio Z-Score M SD M SD
Nonautism Group (n=51) Ratio Z-Score M SD M SD
Domain CA derived predictions Communication 0.449 0.232 −5.837 2.468 0.727 0.164 −2.884 1.742 0.471 0.204 −4.204 1.629 0.663 0.171 −2.672 1.370 Daily Living Skills Socialization 0.384 0.154 −4.697 1.172 0.672 0.132 −2.493 1.009 MA derived predictions Communication 0.613 0.252 −3.813 2.497 0.906 0.168 −0.906 1.664 Daily Living Skills 0.660 0.211 −2.307 1.487 0.848 0.250 −0.986 1.762 Socialization 0.514 0.170 −3.664 1.291 0.835 0.185 −1.236 1.402 aT tests performed on CA derived Z scores: communication, t=−6.64, df=75.9, p<0.01; daily living skills, t=−4.919, df=84.4, p<0.01; socialization, t=−9.744, df=85.5, p<0.01. T tests performed on MA derived Z scores: communication, t=−6.57, df=73.0, p<0.01; daily living skills, t=−3.96, df=93.0, p<0.01; socialization, t=−8.78, df=92.6, p<0.01. A CA-based socialization Z-score cut point of less than or equal to versus greater than −3.5 (for the ratio of actual scores to scores predicted based on CA) most robustly differentiated the autism and nonautism groups. This cut point yielded a sensitivity of .864 and specificity of .863, χ2(1)=49.99, p< 0.01, and resulted in correct classification for 86.3% of cases (Kse=0.741, Ksp=0.710). In previous research (Volkmar et al., 1993), a CA-based cut point of ≤−3.2 versus >−3.2 most robustly differentiated autism and nonautism groups. In the present sample, this cut point yielded a sensitivity of 0.886 and specificity of .784, χ2(1)=42.62, p<0.01, and resulted in correct classification for 83.2% of cases (Kse=0.760, Ksp=0.590). Among the MA-based socialization Z-score cut points, a cut point of less than or equal versus greater than −2.8 best differentiated the autism and nonautism groups. This cut point yielded a sensitivity of 0.795 and specificity of .824, χ2(1)=36.40, p<0.01, and resulted in correct classification of 81.0% of cases (Kse=0.619, Ksp=0.619). In previous research (Volkmar et al., 1993), an MA-based cut point of ≤−2.2 versus >−2.2 best differentiated autism and nonautism groups. This cut point yielded a sensitivity of .886 and specificity of 0.706, χ2(1)=33.77, p<0.01, in the present sample, and resulted in correct classification for 78.9% of cases (Kse=0.737, Ksp=0.483). Adding information about maladaptive Part 2 cut points to (CA- or MAbased) socialization cut points did not significantly increase the number of cases correctly identified.
Annual progress in child psychiatry and child development 2000–2001
342
DISCUSSION The findings of the present study are consistent with several studies demonstrating that individuals with autism display deficits in communication and socialization relative to their peers (Ando et al., 1980; Freeman et al., 1988; Jacobson & Ackerman, 1990; Loveland & Kelley, 1988, 1991; Rodrigue et al., 1991; Schatz & Hamdan-Allen, 1995; Sloan & Marcus, 1981; Volkmar et al., 1987). In our sample, children with autism also displayed significantly poorer daily living skills and more serious maladaptive behaviors than those with other developmental disorders. These differences remained even after controlling for differences in mental age between the groups. Stepwise discriminant function analyses indicated that children with autism could be distinguished from those with other developmental disorders based on deficits in the adaptive behavior domains of socialization and daily living skills as well as on the number of serious maladaptive behaviors exhib-ited. Although both adaptive and maladaptive behaviors distinguished the groups, the failure to make expected gains in adaptive skills was more strongly related to classification than the presence of unusual or serious behaviors. Delays in socialization skills were most strongly related to clinical diagnosis and alone accounted for 48% of the variance in classification. In contrast, serious maladaptive behaviors accounted for only approximately 20% of the variance in classification. These results are consistent with the argument that an impairment in socialization is more central to autism than the presence of unusual or deviant behaviors. This finding is consistent with Kanner’s original description of autism and with research indicating that maladaptive behaviors are displayed by many nonautistic individuals with cognitive and/or sensory delays (Dykens, Hodapp, Walsh, & Nash, 1992; Matson, Baglio, Smiroldo, & Hamilton, 1996; Sparrow et al., 1984). As deviant behaviors (e.g., rocking, self-stimulatory behavior, and self-injurious behavior) are often more apparent to observers than are delays in adaptive skills, these findings have important implications for the assessment of autism. Children with autism exhibited substantial delays in socialization relative to their CA and also relative to their MA. Signal detection analyses indicated that ratios of actual to CA-predicted socialization scores 3.5 or more standard deviations below the normative sample mean most robustly discriminated the autism and nonautism groups. Ratios of actual to MA-predicted socialization scores 2.8 or more standard deviations below the normative sample mean also readily discriminated the two groups. These cut points yielded correct classification for 86.3% and 81% of cases, respectively. These findings indicate that standardized measures of adaptive behavior, such as the Vineland, may be useful in screening for autism. Further studies with larger samples are needed to identify more precisely the best socialization cut points for discriminating autism from other developmental disorders. The identification of these cut points could be useful, because IQ tests and the Vineland Adaptive Behavior Scales are frequently part of clinic and school assessment procedures for children with autism and other developmental disorders. Our findings also suggest that MA-based ratio scores may be more useful in screening for autism as they control for group differences in mental age.
Toward a developmental operational definition of autism
343
Note that the regression method utilized in signal detection analyses in this study and in our previous research is limited to children whose chronological and mental ages fall within the ranges for the Vineland and K-ABC overlap standardization sample (children between the ages of 2.5 and 13 years, and with mental ages between 2.25 and 15.75). Many children with autism have mental ages below the minimum of 2.25. Research is needed that provides normative data on both mental age and adaptive skills in individuals with profound delays in cognitive abilities and adaptive skills. An important area for future research is to compare measures such as the Vineland with recently developed diagnostic instruments, such as the Autism Diagnostic Interview Revised (ADI-R; Lord, Rutter, & Le Couteur, 1994) and the Pre Linguistic Autism Diagnostic Observation Schedule (PL-ADOS; DiLavore, Lord, & Rutter, 1995). The ADI-R and PL-ADOS have been found to reliably distinguish children with autism from those with non-PDD spectrum developmental disorders (DiLavore Lord, & Rutter, 1995; Lord et al., 1997). In contrast to the Vineland, however, these instruments require extensive training and direct assessments of the child. Our findings suggest that applying the regression method to Vineland and IQ scores may provide a useful screener for autism. Children who score below appropriate cut points could then be referred for a more extensive diagnostic evaluation. One difference between our study and previous research was the inclusion of children with PDD-NOS in our sample. We expected that these children would be distinguished from autism and other developmentally disordered children in our sample based on adaptive and maladaptive behaviors. The PDDNOS group differed from the autism group on all variables assessed. Children with PDD-NOS displayed significantly better communication, daily living skills, and socialization than the children with autism. The PDD-NOS group also exhibited fewer maladaptive behaviors than the autism group. These findings are not surprising as PDD-NOS is often considered to be at the less impaired end of a continuum that includes autism. Contrary to predictions, however, children with PDD-NOS could not be distinguished from children with other developmental disorders. The PDD-NOS group did not differ significantly from the nonPDD developmentally disordered group in maladaptive behaviors or on any of the adaptive domains assessed, including socialization. In addition, a multivariate combination of adaptive and maladaptive scores was not successful in distinguishing the two groups. This finding suggests that the Vineland may be of limited utility in screening for PDDNOS. However, it is important to note that the non-PDD developmentally disordered group included in this study was somewhat unusual in that all children had been referred to a clinic specializing in diagnosis and assessment of children with autism and pervasive developmental disorders. Most children were brought to the clinic because a parent, teacher, pediatrician, or other professional raised concerns about the possibility of an autistic spectrum disorder. Thus, although children in the non-PDD developmentally disordered group did not meet criteria for a diagnosis of PDD-NOS, they may have exhibited some of the features of this disorder. This explanation is incomplete, however, because clinicians were able to reliably distinguish these two groups of children within the clinic. A second possible explanation for our failure to identify the PDD-NOS group is that the Vineland domain scores may not be sensitive to more subtle delays or deviations in socialization skills exhibited by children with PDD-
Annual progress in child psychiatry and child development 2000–2001
344
NOS. Further research is needed to determine whether specific items on the Vineland can discriminate children with PDD-NOS from those with other developmental disorders. Measures that assess a wider range of behaviors that are specific to pervasive developmental disorders, such as the Autism Behavior Checklist (Krug, Arick, & Almond, 1980) or the Child Autism Rat-ing Scale (Schopler, Reichler, & Renner, 1988) may prove more useful in screening for PDD-NOS. It is important for future research to explore whether children with PDD-NOS as a group can reliably be distinguished from children with other developmental disorders. Finally, PDD-NOS is an extremely heterogeneous diagnostic category. For some children, the impairments are profound, whereas for others they are much more subtle. Attempts to further subdivide the PDDNOS spectrum (e.g., Klin, Mayes, Volkmar, & Cohen, 1995) may lead to better detection of these children. Recent research on the genetics of autism suggests that autism and PDDNOS reflect different points along a continuum of disability (Baily et al., 1996). The question of the “broader autistic phenotype” remains an open and important one. Note, however, that in our data the operationalization of diagnostic thresholds for screening was very efficient in distinguishing children with autism from those with PDD; this has not been true with other approaches which may tend to overdiagnose autism (Lord, 1997; Lord et al., 1997). In summary, we suggest that evaluating children’s socialization in the context of their developmental level may have advantages over traditional categorical approaches to diagnosis. Such an approach may allow for a better appreciation and understanding of the way in which symptoms change with age. Because this approach relies on standardized and valid assessment tools, like the Vineland, it may lead to more accurate and more reliable diagnoses than traditional categorical approaches. This may reduce confusion about diagnosis among clinicians and parents, and may facilitate research on the prognosis and treatment of this disorder.
ACKNOWLEDGMENT We express our appreciation to the three anonymous reviewers of an earlier draft of this manuscript for their insightful comments and suggestions.
REFERENCES American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., Rev.) Washington, DC: Author. American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.) Washington, DC: Author. Ando, H., Yoshimura, I., & Wakabayashi, S. (1980). Effects of age on adaptive behavior levels in autistic and mentally retarded children. Journal of Autism and Developmental Disorders, 10, 173–183.
Toward a developmental operational definition of autism
345
Baily, A., Phillips, W., & Rutter, M. (1996). Autism: Towards an integration of clini-cal, genetic, neuropsychological, and neurological perspectives. Journal of Child Psychology and Psychiatry, 37, 89–126. Bishop, D.V.M. (1989). Autism, Asperger’s syndrome and semantic-pragmatic disorder: Where are the boundaries? British Journal of Disorders of Communication, 24, 107– 121. Bryson, S.E., Clark, B.S., & Smith, I.M. (1988). First report of a Canadian epidemiological study of autistic syndromes. Journal of Child Psychology and Psychiatry, 29, 433–445. Burd, L., Fisher, W., & Kerbeshian, J. (1987). A prevalence study of pervasive developmental disorders in North Dakota. Journal of the American Academy of Child and Adolescent Psychiatry, 26, 700–703. Carter, A.S., Gillham, J.E., Sparrow, S.S., & Volkmar, F.R. (1996). Adaptive behavior in autism. Child and Adolescent Psychiatry Clinics of North America, 5, 945–961. Carter, A.S., Volkmar, F.R., Sparrow, S.S. Wang, J.J., Lord, C., Dawson, G., Fombonne, E., Loveland, K., Mesibov, G., & Shopler, E. (1998). The Vineland Adaptive Behavior Scales: Supplementary norms for individuals with autism. Journal of Autism and Developmental Disorders, 28, 287–302. DiLavore, P.C., Lord, C., & Rutter, M. (1995). Pre-linguistic autism diagnostic observation schedule. Journal of Autism and Developmental Disorders, 25, 355–379. Dykens, E.M., Hodapp, R.M., Walsh, K., & Nash, L.J. (1992). Adaptive and maladaptive behavior in Prader-Willi syndrome. Journal of the American Academy of Child & Adolescent Psychiatry, 31, 1131–1136. Finucci, J.M., Isaacs, S.D., Whitehouse, C.C. & Childs, B. (1982). Empirical validation of reading and spelling quotients. Developmental Medicine and Child Neurology, 24, 733–744. Finucci, J.M., Whitehouse, C.C., Isaacs, S. & Childs, B. (1982). Derivation and validation of a quantitative definition of specific reading disability for adults. Developmental Medicine and Child Neurology, 24, 733–744. Fombonne, E., & du Mazaubrun, C. (1992). Prevalence of infantile autism in four French regions. Social Psychiatry and Psychiatric Epidemiology, 27, 203–210. Freeman, B.J., Ritvo, E.R., Yokota, A., Childs, J., & Pollard, J. (1988). WISC-R and Vineland Adaptive Behavior Scale scores in autistic children. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 428–429. Greenspan, S.I. (1990). Comprehensive clinical approaches to infants and their families: Psychodynamic and developmental perspectives. In S.J.Meishels & J.P. Shonkoff (Eds). Handbook of early intervention (pp. 150–172). New York: Cambridge University Press. Jacobson, J.W., & Ackerman, L.J. (1990). Differences in adaptive functioning among people with autism or mental retardation. Journal of Autism and Developmental Disorders, 20, 205–219. Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217–250. Kaufman, A.S., & Kaufman, N.L. (1983). Kaufman Assessment Battery for Children. Circle Pines, MN: American Guidance Service. Klin, A., Lang, J., Cicchetti, D.V., & Volkmar, F.R. (in press). Reliability of clinicians in
Annual progress in child psychiatry and child development 2000–2001
346
diagnosing autism: Results of the DSM-IV field trial. Journal of Autism and Developmental Disorders, 30, 163–167. Klin, A., Mayes, L. C., Volkmar, F.R., & Cohen, D.J. (1995). Multiplex developmental disorder. Journal of Developmental and Behavioural Pediatrics, 16, S7–S11. Kraemer, H.C. (1988). Assessment of 2×2 associations and generalization of signal detection methodology. American Statistician, 42, 37–49. Krug, D.A., Arick, J., & Almond, P. (1980). Behavior checklist for identifying severely handicapped individuals with high levels of autistic behavior. Journal of Child Psychology and Psychiatry, 21, 221–229. Lord, C. (1997). Diagnostic instruments in autism spectrum disorders. In D.J.Cohen & F.R.Volkmar (Eds.), Handbook of autism and pervasive developmental disorders (2nd ed., pp. 195–225). New York: Wiley. Lord, C., Pickles, A., McLennan, J., Rutter, M., et al. (1997). Diagnosing autism: Analyses from data from the Autism Diagnostic Interview. Journal of Autism and Developmental Disorders, 27, 501–517. Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism Diagnostic Interview—Revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders, 24, 659–685. Lord, C., & Schopler, E. (1987). Neurobiological implications of sex differences in autism. In E.Schopler & G.Mesibov (Eds.), Neurobiological issues in autism (pp. 191– 212). New York: Plenum Press. Loveland, K.A., & Kelley, M.L. (1988). Development of adaptive behavior in adolescents and young adults with autism and Down Syndrome. American Journal of Mental Retardation, 93, 84–92. Loveland, K.A., & Kelley, M.L. (1991). Development of adaptive behavior in preschoolers with Autism or Down Syndrome. American Journal of Mental Retardation, 96, 13–20. Matson, J.L., Baglio, C.S., Smiroldo, B.B., & Hamilton, M. (1996). Characteristics of autism as assessed by the Diagnostic Assessment for the Severely HandicappedII (DASH-II) . Research in Developmental Disabilities, 17, 135–143. Rodrigue, J.R., Morgan, S.B., & Gefken, G.R. (1991). A comparative evaluation of adaptive behavior in children and adolescents with autism, Down syndrome, and normal development. Journal of Autism and Developmental Disorders, 21, 187–198. Schatz, J., & Hamdan-Allen, G. (1995). Effects of age and IQ on adaptive behavior domains for children with autism. Journal of Autism and Developmental Disorders, 25, 51–60. Schopler, E., Reichler, R.J., & Renner, B.R. (1988). The Child Autism Rating Scale (CARS). Los Angeles, CA: Western Psychiatric Services. Siegel, B., Vukicevic, J., Elliott, G.R., & Kraemer, H.C. (1989). The use of signal detection theory to assess DSM-III-R criteria for autistic disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 542–548. Sloan, J.L., & Marcus, L. (1981). Some findings on the use of the Adaptive Behavior Scale with autistic children. Journal of Autism and Developmental Disorders, 11, 191– 199.
Toward a developmental operational definition of autism
347
Sparrow, S.S., Balla, D., & Cicchetti, D.V. (1984). Vineland Adaptive Behavior Scales (Survey Form). Circle Pines, MN: American Guidance Service. Sroufe, L.A., & Rutter, M. (1984). The domain of developmental psychopathology. Child Development, 55, 17–29. Sverd, J., Sheth, R., Fuss, J., & Levine, J. (1995). Prevalence of pervasive developmental disorder in a sample of psychiatrically hospitalized children and adolescents. Child Psychiatry and Human Development, 25, 221–240. Volkmar, F.R., Carter, A., Sparrow, S.S., & Cicchetti, D.V. (1993). Quantifying social development in autism. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 627–632. Volkmar, F.R., & Cohen, D.J. (1988). Diagnosis of pervasive developmental disorders. In B.Lahey & A.Kazdin (Eds.), Advances in clinical child psychology (Vol. 11, pp. 249– 284). New York: Plenum Press. Volkmar, F.R., Klin, A., Siegel, B., Szatmari, P., Lord, C., Campbell, M., Freeman, B. J., Cicchetti, D.V., Rutter, M., Kline, W., Buitelar, J., Hattab, Y., Fombonne, E., Fuentes, J., Werry, J., Stone, W., Kerbeshian, J., Hoshino, Y., Bregman, J., Loveland, K., Szymanski, L., & Towbin, K. (1994). Field trial for autistic disorder in DSM-IV. American Journal of Psychiatry, 151, 1361–1367. Volkmar, F.R., Sparrow, S.S., Goudreau, D., Cicchetti, D.V, Paul, R., & Cohen, D.J. (1987). Social deficits in autism: An operational approach using the Vineland Adaptive Behavior Scales. Journal of the American Academy of Child and Adolescent Psychiatry, 26, 156–161. Wechsler, D. (1991). Manual for the Wechsler Intelligence Scale for Children-Third Edition (WISC-III). San Antonio, TX: Psychological Corp. World Health Organization. (1993). The ICD-10 classification of mental and behavioral disorders: Diagnostic criteria for research. Geneva: Author. Yule, W., & Rutter, M. (1985). Reading and other learning difficulties. In M.Rutter & L.Hersov (Eds.), Child and adolescent psychiatry: Modern approaches (2nd ed., pp. 444–464). Oxford, UK: Blackwell.
PART IV: OTHER CLINICAL ISSUES
19 Adolescent Onset of the Gender Difference in Lifetime Rates of Major Depression: A Theoretical Model Jill M.Cyranowski, Ellen Frank, Elizabeth Young, and M.Katherine Shear
Prepubescent boys are, if anything, more likely than girls to be depressed. During adolescence, however, a dramatic shift occurs: between the ages of 11 and 13 years, this trend in depression rates is reversed. By 15 years of age, females are approximately twice as likely as males to have experienced an episode of depression, and this gender gap persists for the next 35 to 40 years. We offer a theoretical framework that addresses the timing of this phenomenon. First, we discuss the social and hormonal mechanisms that stimulate affiliative needs for females at puberty. Next, we describe how heightened affiliative need can interact with adolescent transition difficulties to create a depressogenic diathesis as at-risk females reach puberty. This gender-linked vulnerability explains why adolescent females are more likely than males to become depressed when faced with negative life events and, particularly, life events with interpersonal consequences. Arch Gen Psychiatry 2000; 57:21–27
INTRODUCTION Women are about twice as likely as men to experience a lifetime episode of major depression—a finding often described as the most robust in all of psychiatric epidemiology. The female predominance in major depressive disorder is a welldocumented and cross-cultural phenomenon.1,2 Although gender differences in depression once were suspected to be artifactual (representing women’s tendency to report, recall, or seek help for depressive symptoms more than men), artifact theories have largely been discounted as adequate explanations of gender differences in community samples.2 Two large-scale studies conducted in the United States, the Epidemiological Catchment Area Study and the National Comorbidity Survey, have obtained female-male lifetime risk ratios of 2.4:1 and 1.7:1, respectively.3–5 Moreover, data compiled by the Cross National Collaborative Group indicate that the gender difference is apparent across the globe, with female-male ratios ranging from 1.6 in
Annual progress in child psychiatry and child development 2000–2001
350
Beirut, Lebanon, and Taiwan to 3.5 in Munich, Germany.2 Prepubescent girls and boys either do not differ in rates of depression6,7 or, when childhood differences are obtained, boys are more likely than girls to be depressed.8,9 The onset of the gender gap in depression can be traced to a dramatic shift that occurs sometime between the ages of 11 and 13 years, when a precipitous rise in depression rates for adolescent girls far exceeds the modest (if existent) increase displayed by adolescent boys.7,10–12 By 15 years of age, females are twice as likely as males to have experienced a major depressive episode. Although absolute rates of depression continue to rise for both men and women throughout adulthood, the relative predominance of depression in women essentially does not change for the next 35 to 40 years.3,4,13 Several theories have been proffered to explain the adolescent emergence of the gender difference in depression. For example, changes in circulating gonadal hormones during puberty are often implicated as exerting direct or potentiating effects on the central nervous system that relate to disturbances in mood.14 Gonadal hormones are also associated with profound morphological changes that occur during puberty. Some have argued that pubertal changes in female morphology (e.g., breast development and increased body fat) may be experienced negatively by some adolescent girls, particularly if the timing of these changes occurs before that of their peer group.15 Two recent studies, however, indicate that pubertal status has a greater influence on female depression rates than either age10,16 or timing of puberty.10 Brooks-Gunn and Warren17 argued that the “turning on” of the endocrine system as females progress from prepuberty to postpuberty might explain gender-linked increases in negative affect. Accordingly, they found that self-reported negative affect increased in 10- to 14-year-old females during pubertal estrogen level rise. However, estrogen levels accounted for only 4% of the variance in negative affect, whereas negative life events and the interaction of negative life events and estrogen change accounted for 17% of the variance in depressive affect.17 Hence, gonadal hormone change does not seem to be the whole story behind the onset of the gender gap in depression. In addition to biological changes associated with puberty, adolescents undergo equally dramatic transitions in social roles, including major changes in school environments as well as parental, peer, and romantic/sexual relationships. The role of psychosocial factors in depression onset and recurrence is well established in the adult literature, with considerable evidence to suggest that negative life events and chronic psychosocial difficulties place individuals in general, and women in particular, at risk for major depressive episodes.18–20 Recent data suggest that pubertal maturation “sensitizes” females to the depressogenic effects of negative life events. Angold et al21 examined crosssequential data with 1,072 children aged 9 to 16 years, and obtained similarly moderate correlations between life stress and depression for prepubertal boys and girls. However, although the stressdepression relationship became stronger for females as they matured physically, it actually decreased for developing males. Hence, by late puberty boys no longer showed a significant correlation between life stress and depressive symptoms, in contrast to the significant and strong relationship displayed by girls in late puberty. How might pubertal maturation sensitize females to the depressogenic effects of negative life events? We provide a theoretical framework to explain this phenomenon.
Adolescent onset of the gender difference in lifetime rates of major depression
351
First, we describe gender-linked differences in affiliative relationship proclivities and discuss potential social and biological mechanisms that may increase female affiliative needs at puberty. Next, we discuss how pubertal increases in affiliative need interact with adolescent transition difficulty to create a depressogenic diathesis for at-risk females. Finally, we provide evidence to support the increased vulnerability of postpubertal females to depression when faced with stressful life events, and particularly those events with negative interpersonal consequences.
GENDER-LINKED DIFFERENCES IN AFFILIATIVE RELATIONSHIP PATTERNS Females tend to display a strong affiliative style in their social relationships—a tendency that is often attributed to their biologically driven and/or socially prescribed roles as primary child caregivers. By “affiliative style” we refer to a preference for close emotional communication, intimacy, and responsiveness within interpersonal relationships. This preference for affiliation is often set in contrast to a preference for independent activity, mastery, or agency. Indeed, a longstanding heuristic concept in the psychological study of gender differences is the dichotomy between “communion versus agency.”22 This is based on the premise that women are socialized to focus on relationship intimacy or communion, and thereby to develop a sense of self in connection with others. In contrast, males are socialized to focus more on issues of personal autonomy, instrumentality, or agency.23–25 A recent meta-analytic review of more than five decades of personality research supports the existence of this gender-linked dichotomy both over time and across cultures.26 This sex difference in affiliative preference is likely the result of both social and biological factors that evolved from different sexual-reproductive challenges faced by males and females during the course of human evolution—such as women’s historically greater investment in offspring care and their relatively greater use of long-term sexual mate selection strategies (e.g., see research by Buss et al).27,28 Because the evolutionary value of adult affiliative preferences relate largely to sexual and reproductive tasks (such as mating and offspring care), affiliation-enhancing social and biological factors that emerge or intensify during the pubertal transition to reproductive fertility should carry particular value. In the following sections we describe two such processes.
GENDER SOCIALIZATION AND INTENSIFICATION Gender-linked differences in affiliative style are apparent well before puberty,29 as evident in the social interaction patterns displayed in all-girl versus all-boy childhood play groups. A review of psycholinguistic patterns in gender-segregated peer groups30 shows that boys are more likely to command, threaten, and interrupt one another within same-sex group interactions. In contrast, girls are more likely to express agreement, to acknowledge another’s point, and to pause to let another speak. Hence, as Maccoby29 suggests, “among boys, speech serves largely egoistic functions and is used to establish
Annual progress in child psychiatry and child development 2000–2001
352
and protect an individual’s turf. Among girls, conversation is a more socially binding process.”29 Social pressures to conform to stereotypically feminine vs masculine gender roles become even more salient during the pubertal/adolescent transition. This process, referred to as gender intensification,31 is thought to occur for a variety of reasons. Pubertal changes in morphology that make adolescents look more like adults, in addition to increased cross-sex interaction and dating, lead to changes in how adolescents are treated in their social spheres. This may foster changes in adolescent self-perceptions, as well as in how adolescents organize their personal and social activities. Larson and Richards32 have obtained empirical support for the gender intensification hypothesis. Using cross-sectional data from adolescents instructed to record random samples of daily activities, they found that, whereas girls were more social than boys before adolescence, this affiliative orientation intensified during adolescence. Boys increased time spent in solitary activities during the adolescent transition. In contrast, adolescent girls spent less time alone and more time in social activities with peers.33 This emphasis on close interpersonal connection is perhaps most striking in the conversational patterns of adolescent girls. Not only did girls spend more time talking than their male counterparts, but, with increasing age, their conversations showed an increasing interpersonal focus.34 Might there exist parallel, interactive, or actually causative biological processes underlying this intensification of affiliative behavior seen in adolescent females? Research on affiliative behaviors in nonhuman mammals suggests that this may be the case.
THE NEUROBIOLOGY OF MAMMALIAN AFFILIATIVE BEHAVIORS Recent years have seen exponential growth in our understanding of potential neurobiological bases of affiliative behaviors across various mammalian species. Much of this research has highlighted the facilitative role of the mammalian neuropeptide oxytocin (OT). A hypothalamic neurohormone, OT is known to stimulate such mammalian functions as milk ejection during lactation and uterine contraction at parturition. Research suggests that OT also plays a key role in a variety of mammalian affiliative behaviors. For example, OT facilitates the onset of maternal caregiving behaviors in rats and sheep,35–37 and OT released with sexual activity has been shown to foster the development of selective pair-bonding with sexual partners in certain monogamous mammals.38–40 Animal research also indicates potential sex differences in the neurohormonal regulation of affiliative behaviors. Studies on the prairie vole suggest that OT may be particularly relevant to the development of female affiliative behaviors (such as maternal caregiving),38,40,41 a finding with intuitive appeal given OT’s role in female lactational processes (see Carter and Altemus42 for a review). In addition, OT neurotransmission seems to be regulated by fluctuating levels of the female gonadal hormones estrogen and progesterone.41,43–45 Hence, in other mammals, postpubertal changes in female affiliative behavior seem to
Adolescent onset of the gender difference in lifetime rates of major depression
353
involve the OT system, which is regulated by female gonadal hormones. Clearly, we need a more precise understanding of the role of these neurohormonal systems in human affiliative behavior. However, increases in OT activity triggered by female pubertal hormones would fit with the data regarding gender-linked differences in affiliation that become increasingly apparent at adolescence. Such a “biological readiness” for pubertal females to form close affiliative bonds with offspring would have shown clear adaptive value in our premodern ancestors, for whom onset of puberty and onset of child-bearing showed a closer temporal fit. Today, however, puberty occurs earlier46 and childbearing later; hence modern females commonly face a 10- to 15-year lag between the onset of puberty and the onset of childbearing. A pubertal increase in affiliative proclivity should, in itself, have no direct effect on mood. However, for some adolescent females, this increase in affiliative need may occur at a particularly inopportune time, in terms of adolescent transitional difficulties.
THE ADOLESCENT TRANSITION AND THE AT-RISK FEMALE One of the defining transitions to occur during adolescence is the shift from parental to peer/romantic attachment systems. Multiple attachments are thought to be ordered hierarchically, with the primary attachment figure at the top of the attachment hierarchy.47,48 Throughout childhood, parental caregivers tend to serve this primary attachment role. During adolescence, however, a gradual “reshuffling” of the hierarchy occurs. Adolescents increasingly affiliate with peers rather than parents and, by late adolescence or early adulthood, a romantic partner will likely ascend to the role of primary attachment figure.47–49 Most females are able to negotiate the adolescent transition to reciprocal peer relationships and more independent adult roles without undue distress or depression. For a subset of at-risk females, however, this normative transition is experienced as a particularly threatening and difficult one. Here, we briefly describe three risk factors that may give rise to adolescent transition difficulties. Insecure Parental Attachments Adolescents who report an insecure attachment style tend to display lower levels of selfesteem, greater symptoms of psychological distress, greater use of dysfunctional emotion regulation strategies, and lower levels of perceived social support, as compared with their securely attached counterparts.50–54 More specifically, perceived attachment to parents functions to buffer the negative effects of transitional stressors.50,55–57 For securely attached females, use of parental relationships as a “secure base” from which to explore new peer relationships may serve to bridge affiliative needs across the pubertal transition and, subsequently, to buffer the effects of negative life events such as breaches or disappointments in newly formed peer relationships. Indeed, adolescent girls with positive parental relationships are less vulnerable to the depressogenic effects of negative life events.58,59 In contrast, females with strained or insecure parental attachments may form primary peer or romantic-sexual attachments prematurely.48,56 Such attachments
Annual progress in child psychiatry and child development 2000–2001
354
may be both intense and unstable in nature; thus, these relationships may provide less consistent emotional support in the face of negative life events and may, in fact, serve as a direct source of interpersonal life stress. Anxious or Inhibited Temperament Individual differences in the ability to engage socially with others are determined in part by genetic or biologically based temperamental dispositions. In both nonhuman primates60,61 and humans,62–65 a small subset of at-risk infants have been found to display highly reactive, anxious, or inhibited temperamental dispositions that are evident early in life and are, to some degree, heritable.66,67 This inhibited infant profile can increase risk for the development of insecure attachment styles68 and anxiety disorders69 later in childhood. Hence, an anxious or inhibited temperamental profile may both increase stress reac-tivity and interfere with the development of supportive interpersonal relationships. Low Instrumental Coping Skills Females are socialized from an early age to focus on interpersonal affiliation; hence, adolescent females may have less experience relying on independent instrumental coping skills to handle life stressors outside of the context of emotionally close relationships. Research on gender differences in instrumentality support this contention. NolenHoeksema and Girgus70 note that as early as grade 6 females endorse fewer assertive or instrumental traits than their male peers, and this discrepancy may increase during early adolescence.71 Females also evidence a more ruminative style of coping when distressed, in contrast to males’ more active or distracting responses.72 Both low instrumentality73,74 and high ruminative coping styles75 are related to greater levels of subjective distress and depression.
THE DEPRESSOGENIC DIATHESIS Thus, insecure attachments to parents, an anxious-inhibited temperament, and low instrumental coping skills independently (and interactively) contribute to placing certain females at risk for a difficult adolescent transition. We note that these risk factors are often present prior to puberty, and may be evident in males as well as females. However, the social and hormonal milieu in which these risk factors operate changes differentially for the sexes at puberty. We argue that the pubertal timing of the adolescent transition coincides with biologically and socially mediated increases in affiliative need for adolescent females. For the at-risk female, feelings of threat and anxiety engendered by transition difficulties may lead to an even greater focus on interpersonal affiliation (see Figure 19.1). Thus, the at-risk female may simultaneously face high levels of affiliative need, a threatened sense of attachment security, elevated levels of anxiety, and a weak sense of her ability to cope independently with stressful life events. This interaction comprises a potent diathesis for depression in the face of negative life events and,
Adolescent onset of the gender difference in lifetime rates of major depression
355
specifically, events that include interpersonal conflict or loss.
Figure 19.1. Comprehensive representation of the current theoretical model, explaining how various biological and psychosocial forces may interact to produce increased rates of depression in adolescent females. Our list of risk factors, although representative, is clearly not exhaustive. Moreover, atrisk males, who experience a different social and hormonal pubertal environment (that may, for example, foster independent activity over affiliation), may be susceptible to different types of psychological problems (such as externalizing disorders versus the internalizing disorders76 seen in females). Acknowledging the careful work of Kraemer et al,77 we use the term risk in its generic sense. Clearly, future prospective research will be necessary to verify that these variables indeed represent factors that precede depression onset or to substantiate the use of more specific risk terminology.
NEGATIVE LIFE EVENTS Considerable evidence suggests that negative life events place individuals in general, and women in particular, at risk for major depressive episodes.18–20,78–83 Recently, researchers have begun to examine gender differences in how adolescents experience and respond to negative life events, and whether these may affect depressive outcomes. Several studies indicate that adolescent females report more negative life events than their male peers,84–87 and/or that adolescent females experience greater distress than males when negative events do occur.84,85,87,88
Annual progress in child psychiatry and child development 2000–2001
356
Research by Ge et al.59 provides support for gender-linked differences in the stressdepression relationship across adolescence. Using a cross-sequential design that tracked 374 adolescents during a 4-year period, these researchers found that, whereas preadolescent boys reported more negative life events and depressive symptoms than girls, a reversal occurred at age 13 years. After 13 years of age, females reported significantly more negative life events and depressive symptoms than their male counterparts. Moreover, latent growth curve analyses indicated that the trajectory of girls’ depressed mood after 13 years of age was significantly related to changes in the number of life stressors they experienced, whereas the slope in depression trajectory for boys was not affected by the rate of change in life stress.59 These findings dovetail with the results of Brooks-Gunn and Warren,17 Peterson et al.,58 and Angold et al.21 We have also examined the role of stressful life events in adolescent depression, and preliminary findings lend further support to this model. Our research group has compared results of interview-based assessments of negative life events in depressed (28 females, seven males) and nondepressed (20 females, 15 males) adolescents (mean age, 15 years; age range, 12 to 17 years), for the purpose of developing an adolescent version of the Life Events and Difficulties Scale.79 Once remission was achieved, depressed subjects were queried about events occurring 6 months prior to depression onset; controls were queried about a matched time period. Among depressed females, 71% reported one or more severe, negative life events during the 6 months prior to onset, whereas only 35% of nondepressed females and 40% of nondepressed males reported such events during a matched time period. Consistent with the above findings, only 14% of depressed males reported one or more negative life events during the 6-month preonset period. Finally, of the negative life events preceding depression onset in this adolescent sample, the majority (95%) were coded as being “interpersonal” in nature, using a coding scheme we have reported previously.83
CONCLUSIONS Several caveats and clarifications about the current model are in order. First, we do not mean to imply the existence of a social or biological defect in women. We do not argue that either pubertal hormonal changes or affiliative relationship proclivities are themselves dysfunctional, or that they cause depression in adolescent girls. Rather, the hypothesis we present is based on a model of “correlated consequences,” in which: (a) intensified affiliative need experienced in combination with (b) adolescent transition difficulty may create an increased likelihood that (c) negative life events trigger the onset of depressive episodes (see Figure 19.1). Second, we are not arguing that males have less need for attachment relationships than females. Indeed, there is some evidence to suggest that being married confers greater psychological protection for men,89–91 and, some will assert, greater stress for women,1,92,93 patterns that would fit with women’s greater caregiving proclivity. Rather, we argue that a continuum exists across individuals’ preferred level of interpersonal affiliation, and that both biological and social forces combine to place males and females at subtly different places along this continuum. This genderlinked mismatch in affiliative
Adolescent onset of the gender difference in lifetime rates of major depression
357
preference may also explain the “demand-withdrawal” pattern seen in many distressed femalemale marital dyads.94 Third, this article was developed to explicate a particular theoretical frame-work and was not meant to present a critical review of all available literature or all possible variables that may contribute to the existence or onset of the gender difference in depression (for reviews see Weissman & Klerman;1 Wolk & Weissman;2 NolenHoeksema & Girgus;70 and Bebbington95). We have not provided, for example, a thorough discussion of specific genetic or environmental factors that may differentially affect female depression rates. Genetic variables clearly play a role in depression vulnerability but, until recently, had been discounted as an explanation of the gender difference in depression.96 Interestingly, however, a recent adolescent twin study indicates that adolescent females’ genetic heritability emerges only after pubertal onset (and is absent for prepubertal girls, as well as for prepubertal and postpubertal boys).97 Although several different central biological mechanisms may be operating here, these findings support a biologically based vulnerability first expressed for females at puberty. In addition, our article gives short shrift to “environmental” theories that argue that women are more likely than men to experience adverse life events or chronic difficulties, such as childhood sexual abuse, single parenthood, poor employment opportunity, financial strain, and/or an overload of caregiving responsibilities. The current model does not ignore the role played by negative life events, however, and we agree that women may be more likely than men to experience some (but not other) life events or difficulties. Moreover, our theory fits with the “cost of caring” hypothesis,98,99 which posits that women are more affected by events affecting others because they develop more intimate (or affiliative) interpersonal relationships. The female predominance in depression is clearly not a new phenomenon. Existing evidence would suggest that the gender gap has endured since (at least) the 19th century, although some argue that the gap is narrowing over time. As pointed out in a recent review of this topic, these data are not entirely consistent.2 Moreover, several hypothesized cohort effects on depression prevalence rates may more accurately represent age effects. For example, Bebbington et al100 found that the female predominance in depression effectively disappeared after age 55 years—representing a reduction in female prevalence rates corresponding with an age when most women will have passed through menopause. Although further research at this end of the reproductive spectrum is needed, such findings provide additional evidence that the female prevalence in depression is linked to women’s reproductive years. Finally, further exploration of the onset of the gender difference within various Eastern cultures, where gender socialization practices vary yet biologically based affiliative mechanisms related to offspring care may persist, would help to inform the current model. Clearly, we have focused on the explication of females’ affiliative proclivities and potential depressogenic vulnerabilities. Yet the biological and/or social factors that protect adolescent males from the depressogenic effects of negative life events are in need of future elaboration. Finally, we are not argu-ing that all females experience a heightened vulnerability to negative life events during the adolescent transition that will result in depressive symptoms or a depressive syndrome. Indeed, most females will make the adolescent transition without experiencing clinically significant anxiety or depression.
Annual progress in child psychiatry and child development 2000–2001
358
Drawing from the current model, we predict that the most “protected” females are those who display secure attachments to parents, possess instrumental coping skills, and who do not have an extremely anxious or inhibited temperament. Perhaps the primary advantage of this theoretical framework lies in its heuristic value. The model we present connects multiple fields of inquiry and may serve to promote further programmatic research. Clearly, additional research is needed to assess the role of OT in human affiliative behaviors and to determine if puberty indeed primes the OT system for adolescent females. Moreover, longitudinal research designed to assess prepubertal risk factors, pubertal hormonal change, female affiliative tendencies, and negative life events across time will be necessary. In short, we have begun to delineate the pieces of the puzzle that we think will be important to understand why the gender gap in depression emerges at adolescence. Whether each of the individual pieces fit together as we suggest, and how close we have come to mapping out the overall picture, remain to be seen.
ACKNOWLEDGMENTS Many of the ideas included in this article originated from the work of Dr. Frank while she was a fellow at the Center for Advanced Study in the Behavioral Sciences, Palo Alto, CA. Support for this fellowship was provided by the John D. and Catherine T.Mac Arthur Foundation, New York, NY. Preparation of this article was also supported in part by National Institute of Mental Health grant MH18269 (postdoctoral training fellowship to Dr. Cyranowski) and by the John D. and Catherine T.Mac Arthur Research Network on Psychopathology and Development. We acknowledge the thoughtful insights of Janet A.Amico, MD, Barbara L.Andersen, PhD, Floyd Bloom, MD, Tom Boyce, MD, C.Sue Carter, PhD, William H.Durham, PhD, and Paul Pilkonis, PhD, who commented on this work at various stages in its development.
REFERENCES 1. Weissman MM, Klerman GL. Sex differences and the epidemiology of depression. Arch Gen Psychiatry 1977;34:98–111. 2. Wolk Sl, Weissman MM. Women and depression: an update. In: Oldham J, Riba M, eds. American Psychiatric Press Review of Psychiatry. Washington, DC: American Psychiatric Press; 1995; 14:227–259. 3. Weissman MM, Bruce ML, Leaf PJ, Florio LP, Holzer III C. Affective disorders. In: Robins LN, Regier DA, eds. Psychiatric Disorders in America. New York: The Free Press; 1991:53–80. 4. Kessler RC, McGonagle KA, Swartz M, Blazer DG, Nelson CB. Sex and depression in the National Comorbidity Survey, I: lifetime prevalence, chronicity and recurrence. J Affect Disord 1993; 29:85–96. 5. Kessler RC, McGonagle KA, Nelson CB, Hughes M, Swartz M, Blazer DG. Sex and
Adolescent onset of the gender difference in lifetime rates of major depression
359
depression in the National Comorbidity Survey, II: cohort effects. J Affect Disord 1994; 30:15–26. 6. Kashani JH, McGee RD, Clarkson SE, Anderson JC, Walton LE, Williams S, Silva PA, Robins AJ, Cytryn L, McKnew DH. Depression in a sample of nineyear-old children. Arch Gen Psychiatry 1983; 40:1217–1223. 7. Angold A, Rutter M. Effects of age and pubertal status on depression in a large clinical sample. Dev Psychopathol. 1992; 4:5–28. 8. Anderson JC, Williams S, McGee R, Silva PA. DSM-III disorders in preadolescent children: prevalence in a large sample from the general population. Arch Gen Psychiatry 1987; 44:69–76. 9. Cohen P, Brook JS. Family factors related to the persistence of psychopathology in childhood and adolescence. Psychiatry 1987; 50:332–345. 10. Angold A, Costello EJ, Worthman CM. Puberty and depression: the roles of age, pubertal status and pubertal timing. Psychol Med 1998; 28:51–61. 11. Cohen P, Cohen J, Kasen S.Velez CN, Hartmark C, Johnson J, Rojas M, Brook J. Streuning EL. An epidemiological study of disorders in late childhood and adolescence, I: age-and gender-specific prevalence. J Child Psychol Psychiatry 1993; 34:851–867. 12. McGee R, Feehan M, Williams S, Anderson J. DSM-III disorders from age 11 to age 15 years. J Am Acad Child Adolesc Psychiatry 1992; 31:51–59. 13. Frank E, Young E.Pubertal changes and adolescent challenges: why rates of depression rise precipitously for girls between ages 10 and 15. In: Frank E, ed. Sex, Madness, and Society: Gender and Psychopathology. Washington, DC: American Psychiatric Press, in press. 14. Susman EJ, Nottelmann ED, Inoff-Germain G, Dorn L, Chrousos GP.Hormonal influences on aspects of psychological development during adolescence. J Adolesc Health Care 1987; 8:492–504. 15. Stattin H, Magnusson D. Paths Through Life: Pubertal Maturation in Female Development. Hillsdale, NJ: Erlbaum; 1990. 16. Patton GC, Hibbert ME, Carlin J, Shao Q, Rosier M, Caust J, Bowes G. Menarche and the onset of depression and anxiety in Victoria, Australia. J Epidemiol Community Health 1997; 50:661–666. 17. Brooks-Gunn J, Warren MP. Biological and social contributions to negative affect in young adolescent girls. Child Dev 1989; 60:40–55. 18. Brown GW, Harris TO. The Social Origins of Depression: A Study of Psychiatric Disorder in Women. London: Tavistock; 1978. 19. Lewinsohn P, Hoberman H, Rosenbaum M. A prospective study of risk factors for unipolar depression. J Abnorm Psychol 1988; 97:251–264. 20. Paykel E, Myers J, Dienelt M, Klerman G, Lindenthal J, Pepper M. Life events and depression: a controlled study. Arch Gen Psychiatry 1969; 21:753–760. 21. Angold A, Worthman C, Costello E.Puberty and depression: a longitudinal epidemiological diagnostic study. Presented at Annual Meeting of the American Psychopathological Association; February 27, 1997; New York, NY. 22. Bakan D. The Duality of Human Existance: Isolation and Communion in Western Man. Boston, Mass: Beacon Press; 1966.
Annual progress in child psychiatry and child development 2000–2001
360
23. Baker-Miller J. Toward a New Psychology of Women. 2nd ed. Boston: Beacon Press, 1986. 24. Chodorow N. The Reproduction of Mothering: Psychoanalysis and the Ecology of Gender. Berkeley: University of California Press, 1978. 25. Gilligan C. In a Different Voice: Psychological Theory and Women’s Development. Cambridge, Mass: Harvard University Press, 1982. 26. Feingold A. Gender differences in personality: a meta-analysis. Psychol Bull 1994; 116:429–456. 27. Buss DM, Schmitt DP, Sexual strategies theory: an evolutionary perspective on human mating. Psychol Rev 1993; 100:204–232. 28. Buss DM, Barnes M. Preferences in human mate selection. J Pers Soc Psychol 1986; 50:559–570. 29. Maccoby E. Gender and relationships: a developmental account. Am Psychol 1990; 45:513–520. 30. Maltz DN, Borker RA. A cultural approach to male-female miscommunication In: Gumperz JA, ed. Language and Social Identity. New York: Cambridge University Press, 1983:195–216. 31. Hill JP, Lynch ME. The intensification of gender-related role expectations during early adolescence. In: Brooks-Gunn J, Peterson AC, eds. Girls at Puberty: Biological and Psychological Perspectives. New York: Plenum Press, 1983. 32. Larson R, Richards MH, Introduction: the changing life space of early adolescence. J Youth Adolesc. 1989; 18:501–509. 33. Larson R, Richards MH. Daily companionship in late childhood and early adolescence: changing developmental contexts. Child Dev 1991; 62:284–300. 34. Raffaelli M, Duckett E. “We were just talking…”: conversations in early adolescence. J Youth Adolesc 1989; 18:567–582. 35. Keverne EB, Kendrick K. Oxytocin facilitation of maternal behavior. Ann N Y Acad Sci 1992; 652:83–101. 36. Pedersen CA, Ascher JA, Monroe Y, Prange AJ. Oxytocin induces maternal behavior in virgin female rats. Science 1982; 216:648–649. 37. Pedersen CA, Prange AJ. Induction of maternal behavior in virgin rats after intracerebroventricular administration of oxytocin. Proc Natl Acad Sci USA 1979; 76:6661–6665. 38. Carter CS, DeVries AC, Taymans SE, Roberts RL, Williams JR, Getz LL. Peptides, steroids, and pair bonding. Ann NY Acad Sci 1997; 807:260–271. 39. Insel TR, Young L, Wang Z. Molecular aspects of monogamy. Ann NY Acad Sci 1997; 807:302–316. 40. Insel TR, Hulihan TJ. A gender-specific mechanism for pair bonding: oxytocin and partner preference formation in monogamous voles. Behav Neurosci 1995; 109:782– 789. 41. Insel TR. A neurobiological basis of social attachment. Am J Psychiatry 1997; 154:726–735. 42. Carter CS, Attemus M. Integrative functions of lactational hormones in social behavior and stress management . Ann NY Acad Sci 1997; 807:164–174. 43. Amico JA, Crowley RS, Insel TR, Thomas A, O’Keefe JA. Effect of gonadal steroids
Adolescent onset of the gender difference in lifetime rates of major depression
361
upon hypothalamic oxytocin expression. Adv Exp Med Bio 1995; 395:23–35. 44. Crowley RS, Insel TR, O’Keefe JA, Kim NB, Amico JA, Increased accumulation of oxytocin messenger ribonucleic acid in the hypothalamus of the female rat: induction by long term estradiol and progesterone administration and subsequent progesterone withdrawal. Endocrinology 1995; 136:224–231. 45. Thomas A, Amico JA. Sequential estrogen and progesterone (P) followed by P withdrawal increases the level of oxytocin messanger ribonucleic acid in the hypothalamic paraventricular nucleus of the male rat. Life Sci 1996; 58:1615–20. 46. Eveleth P, Tanner J. Worldwide Variation in Human Growth. New York: Cambridge University Press, 1976. 47. Bowlby J. A Secure Base. New York: Basic Books, 1988. 48. Hazan C, Zeifman D. Sex and the psychological tether. In: Bartholomew K, Perlman D, eds. Advances in Personal Relationships. Vol 5. London: Jessica Kingsley, 1994:151–177. 49. Furman W, Buhrmester D. Age and sex differences in perceptions of networks of personal relationships. Child Dev 1992; 63:103–115. 50. Armsden GC, Greenberg MT. The inventory of parent and peer attachment: individual differences and their relationship to psychological well-being in adolescence. J Youth Adolesc 1987; 16:427–454. 51. Blain MD, Thompson JM, Whiffen VE. Attachment and perceived social support in late adolescence. J Adolesc Res 1993; 8:226–241. 52. Cooper ML, Shaver PR, Collins NL. Attachment styles, emotion regulation, and adjustment in adolescence. J Pers Soc Psychol 1998; 74:1380–1397. 53. Hammen CL, Burge D, Daley SE, Davila J, Paley B, Rudolph KD. Interpersonal attachment cognitions and prediction of symptomatic responses to interpersonal stress. J Abnorm Psycho. 1995; 104:436–443. 54. Kobak R, Sceery A. Attachment in late adolescence: working models, affect regulation, and representations of self and others. Child Dev 1988; 59:135–146. 55. Dekovic M, Meeus W. Peer relations in adolescence: effects of parenting and adolescents’ self-concept. J Adolesc 1997; 20:163–176. 56. Nada Raja S, McGee R, Stanton WR. Perceived attachments to parents and peers and psychological well-being in adolescence. J Youth Adolesc 1992; 21:471–485. 57. Papini DR, Roggman LA, Anderson J. Early-adolescent perceptions of attachment to mother and father: a test of emotional-distancing and buffering hypotheses. J Early Adolesc 1991; 11:258–275. 58. Peterson AC, Sarigiani PA, Kennedy RE. Adolescent depression: why more girls? J Youth Adolesc 1991; 20:247–271. 59. Ge X, Lorenz FO, Conger RD, Elder GH, Simons RL. Trajectories of stressful life events and depressive symtoms during adolescence. Dev Psychol 1994; 30:467–483. 60. Suomi SJ, Adolescent depression and depressive symptoms: insights from longitudinal studies with rhesus monkeys. J Youth Adolesc 1991; 20:273–287. 61. Suomi SJ. Uptight and laid-back monkeys: individual differences in the response to social challenges. In: Brauth S, Hall W, Dooling R, eds. Plasticity of Development. Cambridge, Mass: MIT Press, 1991:27–56. 62. Kagan J, Reznick JS, Snidman N. The physiology and psychology of behavioral
Annual progress in child psychiatry and child development 2000–2001
362
inhibition in children. Child Dev 1987; 58:1459–1473. 63. Kagan J, Reznick JS, Snidman N. Biological basis of childhood shyness. Science 1988; 240:167–171. 64. Kagan J. Temperament and reactions to unfamiliarity. Child Dev 1997; 68:139–143. 65. Schmidt LA, Fox NA. Fear-potentiated startie responses in temperamentally different human infants. Dev Psychobiol 1998; 32:113–120. 66. DiLalla LF, Kagan J, Reznick JS. Genetic etiology of behavioral inhibition among 2year-old children. Infant Behav Dev 1994; 17:405–412. 67. Robinson JL, Kagan J, Reznick JS, Corley R. The heritability of inhibited and uninhibited behavior: a twin study. Dev Psychol 1992; 28:1030–1037. 68. Seifert R, Schiller M, Sameroff AJ, Resnick S, Riordan K. Attachment, maternal sensitivity and infant temperament during the first year of life. Dev Psychol 1996; 32:12–25. 69. Hirshfeld DR, Rosenbaum JF, Biedermen J, Bolduc EA, Faraone SV, Snidman N, Reznick JS, Kagan J. Stable behavioral inhibition and its association with anxiety disorder. J Am Acad Child Adolesc Psychiatry 1992; 31:103–111. 70. Nolen-Hoeksema S, Girgus JS. The emergence of gender differences in depression. Psychol Bull 1994; 115:424–443. 71. Galambos NL, Almeida DM, Petersen AC. Masculinity, femininity, and sex role attitudes in early adolescence: exploring gender intensification. Child Dev 1990; 61:1905–1914. 72. Nolen-Hoeksema S, Morrow J, Fredrickson BL. Response styles and the duration of episodes of depressed mood. J Abnorm Psychol 1993; 102:20–28. 73. Allgood-Merten B, Lewinsohn PM, Hops H. Sex differences and adolescent depression. J Abnorm Psychol 1990; 99:55–63. 74. Craighead LW, Green BJ. The relationship between depression and sex-typed personality characteristics in adolescents. J Youth Adolesc 1989; 18:467–474. 75. Nolen-Hoeksema S. Responses to depression and their effects on the duration of depressive episodes. J Abnorm Psychol 1991; 100:569–582. 76. Hay ward C, Killen JD, Wilson DM, Hammer LD, Litt IR, Kraemer HC, Haydel F, Varady A, Taylor CB. Psychiatric risk associated with early puberty in adolescent girls. J Am Acad Child Adolesc Psychiatry 1997; 36:255–262. 77. Kraemer HC, Kazdin AE, Offord DR, Kessler RC, Jensen PS, Kupfer DJ. Coming to terms with terms of risk. Arch Gen Psychiatry 1997; 54:337–343. 78. Bebbington PE, Brugha T, MacCarthy B, Potter J, Sturt T, Wykes T, Katz R, McGuffin P. The Camberwell Collaborative Depression Study, I, depressed probands: adversity and the form of depression. Br J Psychiatry 1988; 152:754–756. 79. Brown GW, Harris TO. Life Events and Illness. New York: Guilford Press, 1989. 80. Kendler KS, Neale M, Kessler R, Heath A, Eaves L. A twin study of recent life events and difficulties. Arch Gen Psychiatry 1993; 50:789–796. 81. Kendler KS, Kessler RC, Walters EE, MacLean C, Neale MC, Health AC, Eaves, LJ. Stressful life events, genetic liability, and onset of an episode of major depression in women. Am J Psychiatry 1995; 152:833–842. 82. Monroe SM, Depue RA. Life stress and depression. In: Becker J, Kleinman A, eds. Psychosocial Aspects of Depression. Hillsdale, NJ: Erlbaum 1991:101–130.
Adolescent onset of the gender difference in lifetime rates of major depression
363
83. Sherrill JT, Anderson BA, Frank E, Reynolds CF, Tu XM, Patterson D, Ritenour A, Kupfer DJ. Is life stress more likely to provoke depressive episodes in women than in men? Depress Anxiety 1997; 6:95–105. 84. Compas BE, Wagner BM. Psychological stress during adolescence: intrapersonal and interpersonal processes. In: Colten ME, Gore S, eds. Adolescent Stress: Causes and Consequences. New York: Aldine de Gruyter, 1991:67–86. 85. Dornbusch SM, Mont-Reynaud R, Ritter PL, Chen Z, Steinberg L. Stressful events and their correlates among adolescents of diverse backgrounds. In: Colton ME, Gore S, eds. Adolescent Stress: Causes and Consequences. New York: Aldine de Gruyter: 1991:111–130. 86. Larson R, Ham M. Stress and “storm and stress” in early adolescence: the relationship of negative events with dysphoric affect. Dev Psychol 1993; 29:130–140. 87. Siddique CM, D’Arcy C. Adolescence, stress, and psychological well-being. J Youth Adolesc 1984;13:459–473. 88. Simmons RG, Burgeson R, Carlton-Ford S, Blyth DA. The impact of cumulative change early adolescence. Child Dev 1987; 58:1220–1234. 89. Bebbington PE. Marital status and depression: a study of English national admission statistics. Acta Psychiatr Scand 1987; 75:640–650. 90. Bebbington P, Hurry J, Tennant C, Sturt E, Wing JK. The epidemiology of mental disorders in Camberwell. Psychol Med 1981; 11:561–566. 91. Bebbington PE, Tansella M. Gender, marital status and treated affective disorders in South Verona: a case-register study. J Affect Disord 1989; 17:83–91. 92. Gove WR. The relationship between sex roles, marital status, and mental illness. Soc Forces 1972; 51:34–44. 93. Gove WR. Sex, marital status, and mortality. Am J Sociol 1973; 79:45–67. 94. Christensen A, Shenk JL. Communication, conflict, and psychological distance in nondistressed, clinic, and divorcing couples. J Consult Clin Psychol 1991; 59:458–463. 95. Bebbington P. The origins of sex differences in depressive disorder: bridging the gap. Int Rev Psychiatry 1996; 8:259–332. 96. Merikangas KR, Weissman MM, Pauls DL. Genetic Facts in the sex ratio of major depression. Psychol Med 1985; 15:63–69. 97. Silberg J, Pickles A, Rutter M, Hewitt J, Simonoff E, Macs H, Carbonneau R, Murrelle L, Foley D, Eaves L. The influence of genetic factors and life stress on depression among adolescent girls. Arch Gen Psychiatry 1999; 56:225–232. 98. Kessler RC, McLeod JD. Sex differences in vulnerability to undesirable life events. Am Sociol Rev 1984; 49:620–631. 99. Turner RJ, Avison WR. Gender and depression: assessing exposure and vulnerability to life events in a chronically strained population. J Nerv Ment Disord 1989; 177:443– 455. 100. Bebbington PE, Dunn G, Jenkins R, Lewis G, Brugha T, Farrell M, Meltzer H. The influence of age and sex on the prevalence of depressive conditions: report from the National Survey of Psychiatric Morbidity. Psychol Med 1998; 28:9–19.
PART IV: OTHER CLINICAL ISSUES
20 Social/Emotional Intelligence and Midlife Resilience in Schoolboys with Low Tested Intelligence George E.Vaillant and J.Timothy Davis
Seventy-three inner-city boys with a mean IQ of 80 were followed prospectively from age 14 until age 65. Their adult adjustment was compared to a socioeconomically matched sample of 38 boys with a mean IQ of 115. Although childhood social disadvantage did not distinguish the groups with low and high IQs, half of the low-IQ men enjoyed incomes as high and had children as well-educated as did the high-IQ men. These resilient low-IQ men were more likely to be generative, to use mature defenses, and to enjoy warm object relations than the high-IQ group as a whole.
INTRODUCTION In determining life outcome, there are always two components: the cards that one is dealt, and how one plays them. In life span studies of resilience, an unanswered question is how one might brilliantly play out the childhood handicap of a low IQ. Indeed, the natural history of low intelligence has been largely ignored. Thus, investigators could recently wonder “why there is such substantial variation in the extent to which low intelligence is accompanied by social deficits” (Rutter, Simonoff, & Plomin, 1996, p. 519). This question is the subject of the present study. In order to understand resilience to low tested intelligence, it must first be acknowledged that the legitimacy of measuring intelligence by conventional assessment scales is controversial. On the one hand, conventional IQ tests measure only the narrow facet of verbal and performance intelligence that predicts school success. A number of alternative models have also commanded respect; these include Sternberg’s (1988) triarchic practical, analytic, and creative/synthetic model of intelligence; Gardner’s (1993) eight-facet model of multiple intelligence (linguistic, logical-mathematical, spatial, interpersonal, intrapersonal, musical, bodily kinesthetic, and naturalist); and Salovey and Mayer’s (1990) model of “emotional” or social intelligence. Furthermore, the “trait” of IQ has been found to be subject to environmental influence, with context, social class, parental education, prejudice, and English as a second language, each
Social/emotional intelligence and midlife
365
exerting a significant effect on measured IQ. On the other hand, as Sternberg (1997) pointed out, in practice, standard intelligence tests have not really changed in over a century. Conventional IQ tests powerfully predict attained education and school success (Jencks, 1979), which in turn influence social mobility, earning power, and even late-life physical health (Guralnik, Land, Blazer, Fillenbaum, & Branch, 1993). In addition, for most children, remedial education or adoption from deprived into advantaged families can shift tested IQ upward by no more than half a standard deviation (7.5 IQ points); and, over the years much of such environmentally induced improvement can be lost (Detterman & Thompson, 1997). Three questions will be addressed in this study. First, in which areas of adult social attainment does IQ matter—and in which areas does it not matter? Second, to what degree can the association of IQ differences with differences in subsequent social attainment be dismissed as an artifact of choice of measurement instrument—and to what degree is it an artifact of other psychosocial antecedents? Third, was Goleman (1995) correct to suggest that, after midlife, the capacity for warm, flexible, and intelligent human relations (emotional intelligence) is far more important than tested analytic intelligence per se. Expressing this third question more concretely: What proportion of intellectually handicapped youth (IQ 60 to 86, M=80) will, in midlife, be as “useful to society” as intellectually more gifted youth (IQ 109 to 130, M=115)? A challenge, of course, is how to select suitable indices, however arbitrary, of late-life social utility. Because adult social class and occupational status are directly linked to years of education, such indices are unduly influenced by low tested intelligence. Conversely, to focus solely on attainment of “happiness” and loving relationships is to minimize the men’s boyhood intellectual handicap. The capacity to love, after all, seems unlikely to be dependent on tested IQ. Therefore, the two criterion outcome measures chosen to reflect social utility were community building (i.e., generativity) and above-average income (a concrete, if reductionistic, index of the value of a person’s work to society). A third outcome measure, children’s educational attainment, was also assessed in order to challenge the assumption that parental IQ and/or social class may be the major rate-limiting factors in children’s education. Illustrative case examples will be presented in order to make these, admittedly arbitrary, efforts at classification more concrete.
METHOD Sample The sample consisted of 456 economically and educationally disadvantaged Boston youth (born between 1925 and 1932) who were originally studied between 1940 and 1945 at an average age of 14.2 years as a nondelinquent control group for a longitudinal study of delinquency (Glueck & Glueck, 1950, 1968). This study provided multiple measures of IQ, as well as assessments of ethnicity, school achievement, social class, and family history of learning difficulties. Equally important, the study provided prospectively gathered information on the men’s ability to love and work until age 65. Finally, the
Annual progress in child psychiatry and child development 2000–2001
366
sample was homogeneous for potential confounders of tested IQ, such as school system (inner city), race (Caucasian), gender (male), and parental educational disadvantage. The men have been reinterviewed at approximately ages 25, 30, and 47. Since age 45, they have been followed by biennial questionnaires. Measures: Early Life Ratings (Before Age 16) While some childhood ratings were made retrospectively, these ratings were based on data gathered in childhood and made by raters blind to all postadolescent information. The raw data came from interviews with the subject, his parents, and his teacher, and from an extensive search of probation, mental health, and social service records before the youth was 16. To reduce halo effects, the first 11 ratings below were made by eight different raters who were blind to most of the other ratings. (Because ratings were done prior to 1974, Pearson’s r was used to assess rater reliability rather than inter-class correlation.) Parental Social Class. This was based on Hollingshead and Redlich’s (1958) threefactor definition of parental social class—father’s education, residence, and occupation— in which class I would include physicians and senior executives with college degrees, living in expensive suburbs (N=0); class II, minor professionals (N=3); class III, whitecollar workers (N=42); class IV, bluecollar workers (N=268); and class V, unskilled laborers with less than high school education living in tenements (N=143). Assessments were based on a home visit. Intelligence. IQ was assessed in junior high school using an individually administered Wechsler Bellevue Intelligence Test (Wechsler, 1939), a normed and validated precursor to the WAIS. The Stanford Intermediate Reading Test was also administered to each boy. The information was not shared with the boy’s teacher or parents. Childhood Competence. This eight-point scale assessed the extent to which, at entrance to the study, subjects had mastered Erikson’s fourth stage of industry—“doing things beside and with others” (Vaillant & Vaillant, 1981). Scale items were: regular home chores (1 point); adjusting well school socially (1 point) and to school academically; taking IQ into account (2 points); participating in extracurricular jobs (1 point) and recreation with others (1 point); and coping well with problems at home (2 points). Interrater reliability (3 raters) ranged from 0.70 to 0.91. Childhood Environmental Strengths. This 3 to 15 point scale, described in detail elsewhere (Vaillant, 1995), reflected a clinical judgment of childhood environmental strengths. Up to five points were assigned for relationships with mother and with father and for an overall home atmosphere conducive to the “development of a basic trust, autonomy and initiative.” Inter-rater reliability (3 raters) ranged from 0.70 to 0.89. Childhood Environmental Weaknesses. This 50-point scale summed 25 concrete criteria of multiproblem family membership, each assigned 0 to 2 points (Vaillant, 1995). Representative items were: eight or more social-agency contacts, raised for more than 6 months apart from both parents, mother mentally retarded, and father cruel to boy. The presence of ten or more such items defined a multiproblem family. Inter-rater reliability (3 raters) ranged from .91 to .94. Childhood health. Scale range was: 1=good physical health (childhood illnesses only;
Social/emotional intelligence and midlife
367
maximum of 2 reported); 3=minor illnesses (3+ childhood illnesses); 5=severe or prolonged physical illness, disability, or handicapping deformity. Withdrawn Neurotic Childhood. Scale range was: 1=good-natured, outgoing childhood; 3=intermediate, no clear problems; 5=very shy, tics, phobias, enuresis past age eight, asocial, or other noted problems. Truancy. The sum of observed reports of truancy (on a yes/no basis) by four contemporary and one retrospective observer. Unofficial Delinquency (Age 12 to 16). Reports by the boy, his parents, and his teachers were summed for the presence=1 or absence=0 of: (a) juvenile drinking; (b) running away; (c) gambling; (d) truancy; (e) sneaking rides; (f) auto stealing; (g) impulsive stealing; (h) planned stealing; and (i) arson (range 0 to 14) (Sampson & Laub, 1993). Scores of 3 to 14 were categorized as unofficial delinquency. Retarded Relatives. Through a review of Department of Mental Health and socialagency records, the number of relatives with mental retardation (IQ < 70) was identified: 1=none, 2=1, 3=2, 4=multiple relatives. Self-Esteem. After review of all the childhood data: 1=good positive selfesteem, 4=adequate in some areas, not in others, 7=strong generalized feelings of inferiority and low self-esteem. (Rater reliability not obtained.) Other Variables. Glueck and Glueck also noted, on a yes/no basis, if the boy rejected school. Measures: Later-Life Ratings The later-life assessments consisted of 13 items in three categories. The first two were the following self-reported items: Social Dancing (Yes/No; age 25). This assessment was based on whether or not, at the interview at age 25, each man mentioned dancing as one of his recreational activities. Adult Job Status (Age 32). Hollingshead and Redlich’s (1958) occupational classification was used, as follows: 1=lawyer, physician, president of large company, 2=architect, university professor, 3=middle manager, 4=technician, 5=skilled laborer, 6=semiskilled laborer, 7=unskilled laborer. Independent raters, blinded to all information collected before subjects reached age 30, rated the men, primarily on the basis of a semistructured 2-hour interview at age 47, on the following eight scales: Social Class (Age 45 to 50). Same scale as used to assess parental social class (above). Income (Age 45 to 55). Maximum earned income in midlife was calculated in 1977 dollars, and then doubled to approximate 1996 dollars. Alcohol Abuse (Age 20 to 50). DSM-III criteria (American Psychiatric Association, 1980) were used: 1=no abuse; 2=alcohol abuse (social complications of alcohol use and difficulty controlling or pathological pattern of alcohol use); 3=alcohol dependence. Generativity (Age 45 to 50). Based on Erikson’s (1963) model, generativity was defined as taking responsibility for the development of the next generation (e.g., as coach, organization leader, or manager). In addition, such individuals had to have achieved intimacy with a significant other for 10 years, and achieved a stable career identity (Vaillant & Milofsky, 1980). Generativity was highly correlated with other objective criteria such as community service and closeness with adolescent children.
Annual progress in child psychiatry and child development 2000–2001
368
Because the kappa was 0.53, the rating classification of “generative” was the consensus judgment of two clinicians blinded to all data before age 30. Of 23 men classified as generative, independent raters disagreed on six; of 69 men classified as not generative, independent rater disagreed on 14. (Although classified by the consensus of two clinicians, 19 men did not receive two independent ratings.) Estimated GAF (Age 45 to 50; Axis V, DSM-IV) (American Psychiatric Association, 1994). This was assessed by Luborsky’s (1962) health-sickness rating scale, on which the GAF (Endicott, Spitzer, Fleiss, & Cohen, 1976) was based. Interrater reliability for this scale was 0.89. Warm Relationships (Age 45 to 50). This was a 25-point scale summing each individual’s relative success in accomplishing eight different tasks of adult object relations, excluding marriage: one task each reflected enjoyment of children, family of origin, and work mates; three tasks reflected friendship network; and two reflected participation in group activities (Vaillant & Vaillant, 1981). Adaptive Defenses (Age 45 to 50). This nine-point scale was based on review of the subjects’ adult lives from ages 20 to 47 via a two-hour interview at age 47. To quantify clinical impressions, the following procedure was observed: 1 to 9 points were assigned for the tendency to deploy adaptive defenses (sublimation, suppression, anticipation, altruism, and humor), rather than maladaptive defenses (projection, schizoid fantasy, turning against self, acting out, hypochondriasis, and neurotic denial/dissociation). Methods, rationale, and reliability have been described elsewhere (Vaillant, 1992). Death or Disability (Age 60). Mortality status was known for all subjects at age 60. An internist, blind to other data, rated complete physicals of the surviving men and noted if they had irreversible disability from chronic disease. Finally, for the purposes of this investigation, three study groups were established. (For reasons of statistical power, cutting points were selected so that each group size would be roughly equal to 35.) The three groups follow: High-IQ Group. This group (N=38) included all study participants with IQs>108. Low-IQ/Poor-Outcome Group. This group (N=39) included all participants with IQs < 87 who failed to be rated as generative and whose annual income was below the mean of $15,000. Low-IQ/Good-Outcome Group. This group (N=34) included all study participants with IQs<87 whose annual incomes exceeded $15,000 or who met criteria for generativity or both. Statistical Procedure The research strategy was, first, to divide the low-IQ subjects into good outcome (GO) and poor outcome (PO) groups, as defined. Next, the variables assessed at concurrent and previous points in the participants’ lives were examined in an attempt to understand the origins of these divergent outcomes. The group of high-IQ men, used as an additional comparison group, was chosen to investigate how the low-IQ/GO men’s attainments contrasted with those of relatively high-IQ men from the same socioeconomically disadvantaged background. Depending on the measurement of the variables, chi-square, Pearson’s r, Spearman’s
Social/emotional intelligence and midlife
369
rho, or Student’s t-test was used to contrast characteristics of men in the subgroups. Despite multiple comparisons, a full Bonferoni correction would be too conservative; in general, only results significant at the p<0.01 level or better receive comment. All pvalues are two-sided.
RESULTS Attrition By age 50, 349 (77%) of the 456 boys originally selected for the study were still active. Thirty-three men (7%) died before age 50, and 74 (16%) withdrew from the study. The 107 excluded men did not significantly differ from the 349 still in the study in terms of IQ or any of the three measures of adolescent adjustment presented in Table 1. Effect of IQ on Later Adjustment Before age 20, IQ (tested at age 14.2) exerted a powerful influence on the lives of these young men, then still schoolboys. Of the 349 men remaining active in the study until age 60, 73 had childhood Wechsler-Bellevue IQs of < 87 (M=80 ± 7) and 38 had childhood IQs > 108 (M=115 ± 5). Those with the highest IQs in childhood attained three more years of education than the men with the low IQs. The latter were ten times more likely to repeat two or more years of school, and almost three times as likely to reject school (see Table 20.1). The effect of IQ on some facets of later adult adjustment, especially those affected by education, appeared equally profound. The 73 men with the lowest IQs in the study were six times as likely to be unskilled laborers at 32, and far more likely to have incomes below the study mean of $30,000. However, other important measures of mental and physical health—generativity, GAP, alcoholism, and physical health—were not significantly associated with IQ. The two low-IQ groups not only differed significantly on the criterion variables of generativity and income, but also in school mastery and adult mental health. Predictors of Comparison-Group Membership As noted, the 73 men with the lowest childhood IQs were divided into two outcome groups. The GO group was composed of the 34 subjects who, in their late forties, had either been rated as generative by interviewers blind to the men’s childhood IQ (N=28) or had incomes above the mean for the total sample (N=20) or both. The other 39 low-IQ men (the PO group) had neither met criteria for generativity nor earned incomes above the mean.
Annual progress in child psychiatry and child development 2000–2001
370
TABLE 20.1 IQ Effects on Facets of Later Adjustment
Period of Adjustment Adolescent—Affected <10 years of school Rejected school
Low IQ PO (N=39) GO (N=34) 74% 82% 62%
41%**b 53%**b 32%*b
High IQ (N=38) 21%***a 24%***a 5%***a
Repeated 2+ gradesc Adult—Affected Unskilled laborer (age 32) 46% 5%***a 15%**b b Social class IV & V (age 47) 90% 26%***a 50%*** Mean income (age 47, 1996 $) 18K (±7) 33K (±13)***b 36K (±11)a Adult—Unaffected 40% Generative (age 47) 0% 59%***b b 18% GAP 85–100 (age 47) 0% 35%*** Alcohol-dependent (age 60) 21% 15% 26% 29% 25% 29% Disabled or dead (age 60) Note. GO=good outcome; PO=poor outcome. aThe two low-IQ groups combined contrasted to the high-IQ group. bThe two low-IQ groups contrasted to each other. cNine (16%) of the low-IQ group, and none of high-IQ group, were in special classes. *p<0.05; **p<0.01; ***p<0.001 (chi-square). The two low-IQ outcome groups differed very little in multiple measures of “analytic” intelligence (see Table 20.2). They also did not differ significantly in terms of the ten individual Wechsler subtests and in performance IQ minus verbal IQ (not shown). However, the low-IQ/GO men differed profoundly from the high-IQ men in every facet of intelligence. In general, the three groups differed surprisingly little from each other in terms of social antecedents (Table 20.2). With the possible exception of parental education and parental mental retardation, neither of the two low-IQ out-come groups differed from the high-IQ group in terms of childhood psychosocial deprivation. Although the finding was not statistically significant, parental retardation was five times as common among men with low tested IQ as it was among the more intellectually gifted.
Social/emotional intelligence and midlife
371
TABLE 20.2 Childhood Psychosocial Antecedents
Antecendents Intellectual Abilities Stanford Reading IQ Wais IQ Verbal IQ, 5 substests Performance IQ, 5 subtests Grades repeated
Low IQ PO (N=39) GO (N=34) 71 (±12) 79 (±6) 78 (±9) 85 (±9) 1.9 (±1)
High IQ (N=38)
98 (±10)***a 115 (±5)***a 111 (±7)***a 117 (±6)***a 0.5 (±0.7) ***a 9.9 (±2)*c 12 (±4)***a 73 (±11) 81 (±6) 80 (±9) 86 (±8) 1.5 (±1)
Years of education 8.7 (±2) Social Antecedents 9 9 11 Parent’s education (apx grades completed) Mother or father retarded 18% 15% 3% Childhood health poor 33% 38% 24%b Weekly per capita income, 1945 dollars 9.50 (±2.50) 9.25 (±2.40) 9.00 (±2.80) (age 14) Parental social class 4.2 (±0.7) 4.3 (±.5) 4.1 (±7) 9.9 (±3.6) 9.1 (±3.8) 8.9 (±4.2) Environmental strengths Environmental weaknesses 11.9 (±7.5) 10.0 (±5.5) 10.6 (±8.0) Psychological Antecedents Poor childhood competence 28% 9% 11% c 79% Social dancing (Age 25) 56% 82%*** c 11% Poor self-esteem 28% 9%** 44% 24% 29% Unofficial delinquency >3 Note. Age=14 (±2) unless otherwise noted. Outcome figure are means (with SD in parentheses) or percentages. GO=good outcome; PO=poor outcome. aGroups compared were low-IQ/GO men and high-IQ men. bFor the whole sample, poor childhood health correlated negatively with high IQ (r=15, p=.005). cGroups compared were the low-IQ GO and PO men. *p<0.05; **p<0.01***; p<0.001 (Student’s t-test and chi-square). It was in terms of antecedent social skills that the two low-IQ groups differed most. Those in the GO group were far more able to report that they enjoyed social dancing at age 25. (Because of the absence in the Glueck data set of more detailed measures of the various aspects of object relations in young adulthood, social dancing was selected as the best available proxy variable for the presence or absence of social skills. In fact, of all
Annual progress in child psychiatry and child development 2000–2001
372
variables gathered before age 32, social dancing proved to be the best early predictor of warm midlife relationships.) The GO group was a little less likely than the PO group to be emotionally troubled in terms of unofficial delinquency and low self-esteem, and was significantly superior in childhood competence. In adult life, the two low-IQ groups differed markedly from each other in object relations, mental health, and adaptive defenses (see Table 20.3). Indeed, 56% of the men in the GO group manifested predominantly adaptive defenses (especially anticipation, suppression, and altruism), in contrast to almost none of those in the PO group. The men with poor outcomes were more likely to use maladaptive defenses, especially of turning against self, projection, and fantasy. (Raters of defenses were blind to IQ and generativity ratings.)
TABLE 20.3 Adult Social Utility and Emotional Intelligence (Age 47)
Antecendents Outcome Variablesa Mean annual income Generative Independent Variables Social class II to III GAF 80+ Warm relationships Married 15+ yrs. Mature defenses Children, 1+ yrs. of college
Low IQ PO (N=39) GO (N=34) 17K (±7) 0%
38K (±10) 59%
High IQ (N=38) 36K (±14) 40%
74%*c 50%**b 45% 50%**b 45% 53%**b b 84% 97%** b 43% 56%** 66% 56%**b aSig. levels between low-IQ PO and GO not reported because these variables were used to assign subjects to outcome group. bStatistical test between low-IQ PO and GO groups. cSocial class of high-IQ group was significantly higher than low-IQ/GO group. *p<0.05; **p<0.001 (chi-square). 10% 5% 11% 54% 13% 13%
Despite their profound differences in intelligence, the 34 low-IQ/GO men did not differ substantially on any measure (except social class, p<0.05) from the 38 men with tested intelligence an average of 35 points higher. For example, the children of men in the GO group were four times more likely to go to college than were those of the PO group—but just as likely to go to college as were the children of the high-IQ men. Logistic regression, using all the childhood psychosocial variables, including social dancing, showed that only childhood competence and social dancing predicted membership in the GO group. Childhood competence was an amalgam of childhood environment, ego strength, perseverance, and relationship skills. Similarly, social
Social/emotional intelligence and midlife
373
dancing—the age 25 proxy variable—was the best predictor of adult object relations. A competing explanation for these men’s “recovery” from their early low IQ is that the test scores of the GO men were transiently suppressed due to English being a second language. In the present sample, no such association was observed. The only finding that was even suggestive in this respect was that the Stanford Reading Quotient (SRQ) scores of the 197 men with U.S.born parents were modestly higher (86 ±15) than those of that portion of men in the group with parents born in Italy (83±15, f=4.2, p=.04).
CASE HISTORIES In order to make the nature of resilience to low intelligence more vivid, eight of the most dramatic case histories are offered. The importance of both manual and social dexterity to resilience from intellectual deficit is striking. (All names are fictitious and identifying material has been deleted.) Case 1 John Mayer had a performance IQ of 79, and a verbal IQ of 83. His SRQ was 72. As expected, he had only completed nine grades. However, he demonstrated increased social skills from early in his life. He was described at 14 as “conscientious,” “likable, direct” and as having “very good poise [and a] considerable sense of humor.” In his adult life he has enjoyed an excellent marriage and has transformed his love for kids and skill at sports into a $60,000-a-year position as director of parks and recreation in a city of 90,000. For the past 25 years, his city and job have been growing slowly enough for him to adjust to expanding responsibility. His wife has always assisted him by keeping the books. Currently, he has 160 people working for him, and after retirement he plans to run for the job of city commissioner. He loves his job, his wife, and his grandchildren. His five children all have high school diplomas. Case 2 Bill DiMaggio had an IQ of 82 and an SRQ of 71. His father was a laborer who became disabled when Bill was a teenager; his mother died when he was 16. His family lived in a tenement without central heating, where he shared a bed with his brother. He completed ten grades of school with difficulty, and repeated one grade. His teacher wrote, “This boy would spoil any class because of his behavior (impatience, defiance, disorderliness, attracting attention) which may be due to nervousness. Tardy 4 days a week.” However, an interviewer at this time also caught glimpses of DiMaggio’s social skills and saw him as “emotionally stable” and “quite gregarious.” At 31, DiMaggio was described as “mature [and] self-assured” and was praised by his wife as a “good husband” and a “wonderful father.” At 47, he was described by the interviewer (blinded to his past) as a charming, responsible, and committed man, who maintained warm eye contact. After 15 years as a laborer for the Massachusetts Department of Public Works, he was promoted to carpenter through seniority. He derived
Annual progress in child psychiatry and child development 2000–2001
374
great pride in maintaining Boston’s historic municipal buildings. He had rebuilt his own house, and his relations with his wife and children were exceptional. Because he was the union shop steward, management had to listen to him; they also depended on him to teach the other men. DiMaggio was on the board of the Council of Organizations, the umbrella group for all the charities in his urban community. Case 3 Butch Bishop’s Wechsler IQ was 86, his SRQ was 78, and his Math Quotient was 63. As a child, he had suffered Sydenham’s chorea and repeated first and fifth grades. Although he had an abusive, alcoholic father and a seriously delinquent brother, two older siblings and his mother were ardently devoted to the Salvation Army. When Butch was not babysitting his younger sibs, he spent most of his time working for the Salvation Army. In regular school, he received a D for effort and his teacher described him as “slovenly, tardy, and lazy.” In Salvation Army, however, he achieved a perfect 12-year Sunday school attendance record. After leaving school in 11th grade, Butch took four 1-month university summer courses to become a “supply minister” for the Methodist church. (Supply ministers work in parishes too small to pay a trained minister.) With the help of his minister brother-in-law, he got a small parish and has spent the last 34 years in charge of progressively larger ones; his wife runs the business side. He called his work “exciting and rewarding,” and particularly enjoyed “helping and teaching people.” His children have all completed 1 to 2 years of college. He still cannot spell, but he is on the board of directors of the Salvation Army. Although all three of the men described were failures at school, all achieved community leadership roles. Further, all of them received support from their wives. For the five men whose histories are outlined in the following, good marriages also seemed crucial. Case 4 Fred Sergeant had an IQ of 86, repeated three grades in school, and had found just going to school to be the hardest part of growing up. He was the ninth of 12 children; his father was illiterate. He had a knack, however, for finding his niche. In the Army, Fred became a supply sergeant; he supervised nine employees and three departments, and earned $40,000 a year. His obsessional character was well suited to the stockroom. He loved his work and felt it was never boring. Fred married a nurturant older woman who gave him the affection that he had missed in his crowded family. When asked what pleased him most about her, he replied, “The love and comfort she has given me all these years.” Case 5 Perry Rand (IQ of 85) repeated three grades before dropping out of school after ninth grade. Early in their happy marriage, Perry’s wife advised him to re-enlist in the Navy because of better opportunities for men with limited education. He became a Navy IBM machine operator, and later a government computer specialist making $50,000 a year. All
Social/emotional intelligence and midlife
375
four of their children attended college and two graduated. Case 6 Bill Eagle had an IQ of 60 and an SRQ of 58. He was the 13th of 14 children, only six of whom survived. As a child, he carried a diagnosis of “feebleminded,” and he spent his school years in special ungraded classes. He had a cleft palate, lost the sight of his left eye, the hearing in his left ear, and suffered multiple skull fractures. There was mental retardation on both sides of the family, and he had enormous difficulty writing out answers to questionnaires. He spent all his life working at the same loading dock. He enjoyed a happy marriage to his wife and boasted, “She takes care of me and will do anything I ask her to.” Both of his two sons were high school graduates and Eagle Scouts. He had taken an active role in leading their camping trips and in scouting in general. Case 7 John Gamble had an IQ of 73 and an SRQ of 78. He repeated two grades. One teacher described him as, “the worst student I have ever seen, mean and stolid, ready to pounce on anyone. He will calm down when rattaned.” By age 60, Gamble owned his home, worked as a professional card dealer, and had a good marriage. Except as a card dealer, he had no particular skills with his hands. For recreation, he read Civil War books. Two of his children graduated from high school; the other completed two years of college. Case 8 As a child, Bill Carpenter shared a bedroom with two brothers; his apartment had no tub and no hot water. He was close to his father but his mother, who had an estimated IQ of 50, was deemed by the Society for Prevention of Cruelty to Children to be suitable for institutionalization. When he left school in ninth grade, however, he recorded an IQ score of 111 on the Dearborn reading test, and he repeated only one grade in school. His skill at carpentry led the interviewer to describe the interior of his house as “quite amazing.” He earned $40,000 as a post office supervisor, and he saw his job in psychologically minded terms. He was vice-president of his union, coach of his son’s winning football team, and an officer in the Knights of Columbus. He had pioneered getting adolescents in his neighborhood involved in religious education. His marriage of 47 years was the center of his life.
DISCUSSION Intelligence can be narrowly defined by means of tests designed to predict school performance—or intelligence can be broadly defined by assessing the mental abilities necessary to adapt, shape, and select one’s environment (Sternberg, 1997). This paper has focused on the narrow, socially biased construct of intelligence for the following reasons. First, early in life, IQ tests do predict education, school performance, and teacher
Annual progress in child psychiatry and child development 2000–2001
376
approval. In turn, school success and teacher approval are closely linked with resilience in the face of familial deprivation (Quinton & Rutter, 1988). In this sample, IQ, tested when the boys all still had comparable education, predicted future educational attainment (r=.37, p<.001). Furthermore, in predicting education beyond high school for ninth graders, IQ seemed more important than either social class or parental expectation (Rehberg & Rosenthal, 1978). In short, narrowly defined IQ is an important predictor of success, resilience, and self-esteem in youth. Admittedly, in the general population, tested IQ and its consequences are also profoundly affected by social environment. In this study, however, the impact of many relevant facets of social environment—economic opportunity, gender, race, school quality, access to the G.I. bill, and birth cohort were all controlled. The degree to which it was their wives’ intelligence that made a difference in the outcomes of the low-IQ men, especially in their children’s attained education, cannot be calculated from the present data. Certainly, good marriages were important. For example, for the entire sample, if boyhood competence, education, social dancing, and stability of marriage were controlled through multiple regression, then neither parental social class nor IQ made a further contribution to predicting income or generativity. Finally, during the school years, low tested IQ is a terrible curse, but IQ is not destiny. In the first place, IQ is not very stable across multiple generations. In the present sample, the estimated number of retarded relatives correlated only weakly with the men’s IQ; further, the attained education of these men’s children seemed more a function of their father’s capacity for generativity than of his IQ. In a three-generational study of men with mild retardation (IQ 50 to 70), although 16% of the probands’ children and 33% of their parents also manifested IQs below 70, the IQs of only 4% of their grandchildren tested below 70 (Reed & Reed, 1965). Resilience with respect to low tested intelligence appears to be mediated by those social skills often termed emotional intelligence, which Goleman (1995) defined as being able to motivate one’s self and persist in the face of frustrations; to control impulse and delay gratification; and to keep distress from swamping the ability to think, empathize, and hope. Emotional intelligence is manifested by people who know and manage their own feelings well, and who read and deal effectively with other people’s feelings. Such skills are epitomized by the differences in adaptive defenses noted in Table 20.3. In the present study, half the boys with low tested IQ and multiple grade repetitions eventually fared as well as boys with a mean IQ almost two standard deviations higher on variables designed to reflect “value to society”—income, community leadership, and children’s educational achievement. These men who turned out well manifested superior social skills, manual dexterity, and a capacity for rewarding marriages. Reciprocally, the harbingers of failed adult adjustment in men with low IQs were not differences in IQ or education; rather, poor outcome was predicted by poor social and emotional competence in childhood and adulthood. Thus, the present findings suggest that the concept of emotional intelligence is one deserving of considerably more attention, and the research in that area by investigators like Gardner (1993), Salovey and Mayer (1990), and Sternberg and Smith (1985) should be further developed by students of resilience.
Social/emotional intelligence and midlife
377
REFERENCES American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Detterman, D.K., & Thompson, L.A. (1997). What is so special about special education. American Psychologist, 52, 1082–1090. Endicott, J., Spitzer, R.L., Fleiss, J.L., & Cohen, J. (1976). The global assessment scale: A procedure for measuring overall severity of psychiatric disturbance. Archives of General Psychiatry, 33, 766–771. Erickson, E. (1963). Childhood and society (2nd ed.). New York: Norton. Gardner, H. (1993). Multiple intelligences: The theory in practice. New York: Basic Books. Goleman, D. (1995). Emotional intelligence. New York: Bantam Books. Glueck, S., & Glueck, E. (1950). Unraveling juvenile delinquency. New York: Commonwealth Fund. Glueck, S., & Glueck, E. (1968). Delinquents and nondelinquents in perspective. Cambridge, MA: Harvard University Press. Guralnik, J.M., Land, K.C., Blazer, D., Fillenbaum, G.G., & Branch, L.G. (1993). Educational status and active life expectancy among older blacks and whites. New England Journal of Medicine, 329, 110–116. Hollingshead, A.B., & Redlich, F.C. (1958). Social class and mental illness. New York: John Wiley. Jencks, C. (1979). Who gets ahead? The determinants of economic success in America. New York: Basic Books. Luborsky, L. (1962). Clinicians’ judgments of mental health. Archives of General Psychiatry, 7, 407–417. Quinton, D., & Rutter, M. (1988). Parenting breakdown: The making and breaking of intergenerational links. Brookfield, VT: Aldershot. Rehberg, R.A., & Rosenthal, E.R. (1978). Class and merit in the American high school: An assessment of the revisionist and meritocratic arguments. New York: Longman. Reed, E.W., & Reed, S.G. (1965). Mental retardation: A family study. Philadelphia: Saunders. Rutter, M., Simonoff, E., & Plomin, R. (1996). Genetic influences on mild mental retardation: Concepts, findings and research implications. Journal of Bio social Science, 28, 509–526. Salovey, P., & Mayer, J.D. (1990). Emotional intelligence. Imagination, Cognition and Personality, 9, 185–211. Sampson, R.J., & Laub, J.H. (1993). Crime in the making. Cambridge, MA: Harvard University Press . Sternberg, R.J. (1988). The triarchic mind: A new theory of human intelligence. New York: Viking Press.
Annual progress in child psychiatry and child development 2000–2001
378
Sternberg, R.J., & Smith, C. (1985). Social intelligence and decoding skills in nonverbal communication. Social Cognition, 3, 168–192. Sternberg, R.J. (1997). The concept of intelligence and its role in lifelong learning and success. American Psychologist, 52, 1030–1037. Vaillant, G.E. (1992). Ego mechanisms of defense. Washington, DC: American Psychiatric Association Press. Vaillant, G.E. (1995). Natural history of alcoholism revisited. Cambridge, MA: Harvard University Press. Vaillant, G.E., & Milofsky, E.S. (1980). Natural history of male psychological health: IX. Empirical evidence for Erikson’s model of the life cycle. American Journal of Psychiatry, 137, 1348–1359. Vaillant, G.E., & Vaillant, C.O. (1981). Natural history of male psychological health: X. Work as a predictor of positive mental health. American Journal of Psychiatry, 138, 1433–1440. Wechsler, D. (1939). The measurement of adult intelligence. Baltimore: Williams & Wilkins.
Part V TREATMENT ISSUES
PART V: TREATMENT ISSUES
21 Effectiveness of Nonresidential Specialty Mental Health Services for Children and Adolescents in the “Real World” Adrian Angold, E.Jane Costello, Barbara J.Burns, Alaattin Erkanli, and Elizabeth M.Z.Farmer
Objective: Although many studies demonstrate the efficacy of a variety of treatments for child and adolescent psychiatric disorders, studies showing the effectiveness of such treatments in ordinary clinical settings have not been forthcoming. This report presents a study of the effectiveness of outpatient treatment in a community sample of 9- to 16-year-olds. Method: Four annual waves of data were collected from a representative sample of 1,422 children and their parents in the southeastern United States. Interviews were conducted with the Child and Adolescent Psychiatric Assessment to determine clinical status and the Child and Adolescent impact Assessment to measure the impact of psychiatric disorder on the lives of the children’s families. Results: Treated individuals were more severely disturbed and showed deterioration in their clinical status, even before they received treatment, indicating that comparisons with untreated individuals required controls not only for pretreatment clinical status, but for pretreatment clinical trajectory. A significant dose—response relationship was found between the number of specialty mental health treatment sessions received and improvement in symptoms at followup. However, no effect of treatment on secondary psychosocial impairment or parental impact was identified. Conclusions: Child and adolescent outpatient psychiatric treatment has positive effects on psychiatric symptoms, even when conducted outside the academic units where efficacy research usually takes place. The dose of treatment required to produce such effects (more than eight sessions) suggests that attempts to limit child psychiatric treatment to very short-term interventions may be counterproductive. (J Am Acad Child Adolesc Psychiatry, 2000, 39(2): 154–160.) Key Words: children and adolescents, outpatient, treatment.
Effectiveness of nonresidential specialty mental health services
383
INTRODUCTION Most controlled treatment trials (efficacy studies) (U.S. Congress, 1991) take place in resource-rich, university-based settings and involve highly selected, often non-referred subjects who are willing to be randomized and who have relatively homogeneous histories. They often exclude those who are receiving other treatments, and the experimental treatment is carefully controlled to ensure adherence to the protocol (Hoagwood et al., 1995; Seligman, 1995; Weisz & Hawley, 1998). As a result of numerous demonstrations of the efficacy of various forms of psychotherapy, family therapy, and medication in the treatment of child and adolescent psychiatric disorders (Casey et al., 1985; Jensen et al., 1996; Richters et al., 1995; Weisz & Hawley, 1998; Weisz et al., 1987, 1992, 1995), increasing attention is being paid to the question of whether such treatments are effective in the real world of the mental health practitioner’s office (Weisz et al., 1998). In the real world, multiproblem children in poorly resourced environments receive relatively individualized treatment packages that may be driven more by the realities of therapist scheduling, family willingness and ability to participate, and insurance coverage than they are by the niceties of adherence to a prespecified treatment protocol (Seligman, 1995; Weisz & Hawley, 1998). In these circumstances, we have to ask the question, Is treatment effective in the community (Hoagwood et al., 1995)? More than 30 years ago, Shepherd et al. (1966) matched 50 consecutive child guidance clinic attendees with 50 subjects from a general population survey on their clinical profiles and found that there was no difference in clinical outcomes 2 years later; in both groups, most children had improved. Since then, several—mostly small, and with one exception (Weisz et al., 1992) old—studies of psychotherapy as it is delivered in ordinary clinics concluded that there is no evidence for effectiveness (Ashcraft, 1971; DeFries et al., 1964; Levitt, 1957a, 1957b). The most recent and—given recent progress in psychotherapeutic techniques—the most relevant study is that of Weisz and Weiss (1989). In their study of nine child guidance clinics, scores on the Child Behavior Checklist (Achenbach & Edelbrock, 1991) and teacher ratings did not differ at 6 months and 1 year between those who had dropped out of outpatient psychotherapy after only an initial assessment and those who stayed for at least five sessions and then terminated with the therapist’s approval. They concluded that “therapy conducted by working clinicians, in actual clinics, with spontaneously referred children, under everyday conditions, may not be as effective as the therapy conducted under controlled conditions for research purposes” (Weisz & Weiss, 1989, p. 746). This appropriately modest conclusion masks the stark fact that their study found no evidence that the treatment provided did any good at all. This impression has recently been reinforced by the Fort Bragg Demonstration Project (Bickman et al., 1995), in which a greatly enhanced outpatient mental health care system was provided at no cost to families, in an attempt to reduce inpatient care. While access to care without limits on expenditures increased the number of children served and the cost per child, clinical outcomes were no better than those of children seen in a control
Annual progress in child psychiatry and child development 2000–2001
384
setting without the additional resources. When a substantial increase in resources in unaccompanied by any increase in apparent effectiveness, one possible explanation is that neither the experimental nor the control intervention had any effect on the clinical status of the children they served. These studies attempted, in different ways, to bring methodological rigor to bear on testing for treatment effects in real-world clinical settings, but they failed to find a treatment signal above the noise of the many potentially confounding factors. Here we tackle the problem using data from a community study in which repeated assessments permit us to take a nontraditional approach to examining the effectiveness of treatment.
METHOD Setting and Sample The Great Smoky Mountains Study is an ongoing, longitudinal study of the development of psychiatric disorders and need for mental health services in rural and urban youths. Details of the study design and instruments used can be found elsewhere (Costello et al., 1996). Briefly, a representative random sample of children aged 9, 11, and 13 years in 1993 was recruited from 11 rural counties in western North Carolina. American Indian youths, and children with behavioral problems (reported by parents in a brief telephone questionnaire), were oversampled. The overall response rate at the first wave was 80% (N=1,422). Contact with families was maintained by annual interviews and by telephone every 3 months. At the end of 4 years, information was available on the age range 9 through 16, with overlapping data from two age cohorts at ages 11, 12, 13, and 14. In all, the sample included 4,966 annual observations on the 1,422 subjects (87% of potential interviews on those subjects). Thus across four waves, 70% (80%×87%) of all interviews with eligible subjects were completed. However, here we restricted our analyses to the 997 (weighted prevalence=59%) individuals who could be regarded as having potential need for psychiatric services at 1 or more of the 4 waves of assessment, because of meeting DSMIII-R (American Psychiatric Association, 1994) criteria for one (51%) of 29 well-defined diagnoses or having psychiatric symptoms causing significant psychosocial impairment (49%). The latter may be regarded as suffering from DSM-III-R “not otherwise specified” (NOS) disorders (Angold et al., 1999). There were a total of 3,532 observations over the four waves on these children with potential need for treatment. We cannot provide an exact response rate for this group, because the subsetting depends on responses to the interview, so uninterviewed subjects cannot be categorized. However, individuals with high screen scores were not less likely than those with low screen scores to consent to be interviewed (Costello et al., 1996), so it is reasonable to suppose that the interview response rate in the potentially in need subgroup considered here would have been similar to the overall interview response rate. Measures Child and Adolescent Psychiatric Assessment. The Child and Adolescent Psychiatric
Effectiveness of nonresidential specialty mental health services
385
Assessment provides a structured questioning scheme for use with both children and parents or guardians that enables interviewers to determine whether symptoms, as defined in an extensive glossary, are present or absent and to code their frequency, duration, and onset (Angold et al., 1995). Diagnostic 1-week test-retest reliabilities for child selfreports range from 0.55 for conduct disorder to 1.0 for substance abuse/dependence (K) (Angold & Costello, 1995). Diagnoses and symptom scores, including a count of the total number of DSM-III-R psychiatric symptoms relating to any of 29 separate diagnoses, were generated by computerized algorithms which combine information from the child and parent. Psychosocial impairment secondary to psychiatric symptoms in 17 areas of functioning related to life at home, at school, and elsewhere was also rated according to a series of definitions and rules specified in the glossary. The test-retest intraclass correlation coefficient for level of psychosocial impairment by child self-report was 0.77 (Angold & Costello, 1995). Child and Adolescent Impact Assessment. The Child and Adolescent Impact Assessment (Angold et al., 1998; Messer et al., 1996) was completed by parents at the end of the diagnostic interview. Parents were asked about 24 potential perceived “impacts”—that is, problems, difficulties, or burdens in their own lives that they perceived as being caused or exacerbated by their child’s behavioral or emotional problems. The areas covered included expense and financial difficulties; problems in their relationships with their spouse, family, or social network members; restrictions on activities; and decreased feelings of well-being and competence. Child and Adolescent Services Assessment. The Child and Adolescent Services Assessment (CASA) was used to collect parent and child reports on use of mental health services, broadly conceptualized to include services provided both by the specialty mental health sector and by schools, social services, primary health care, juvenile justice, and informal community sources (Ascher et al., 1996; Farmer et al., 1994). To obtain complete service histories, interviewers contacted a parent (usually by telephone) every 3 months between the annual interviews and completed the CASA for the period of the preceding 3 months. The key variable from the CASA used here is the number of outpatient specialty mental health visits reported over a 2-year period. Analysis The follow-up symptom, impairment, and parental impact scores were all substantially skewed to the left, in a reasonable approximation to a Poisson distribution, so Poisson regressions were performed. All analyses were weighted to provide unbiased general population estimates of effects. In addition, “sandwich” variance corrections (Binder, 1983) were used to correct for the effects of the 2-stage sampling scheme.
RESULTS Over the course of the study, subjects received outpatient treatment from a variety of private practitioners of various disciplines (N=189), as well as from public mental health centers (N=135), drug treatment programs (N=5), a crisis center (N=8), and an in-home
Annual progress in child psychiatry and child development 2000–2001
386
treatment program (N=35). In addition, 39 had received psychiatric treatment at a community health center and 53 had received some residential or inpatient services. Because we are concerned in this report with nonresidential mental health treatment, contracts with these latter settings are not considered in the analyses presented here. Controlling for these contacts made no meaningful difference to the results we present (analyses involving these controls are available from the first author upon request).
Figure 21.1. Severity of disorders in the untreated and pretreatment groups. Treated individuals had higher levels of symptoms, impairment, and perceived impact before treatment than individuals in need who did not receive treatment. Figure 21.1 shows the total numbers of DSM-III-R symptoms, levels of impairment, and perceived impact in individuals who did not receive treatment at any point, compared with the pretreatment observations from those who would receive specialty mental health treatment in the future but had not done so yet. Data from wave 4 were necessarily excluded from this analysis, because we did not know who would turn out to be a service user during the following year (no data on treatment subsequent to interview wave 4 being available), so we could not distinguish nonusers from the pretreatment group at wave 4. A total of 2,736 observations were available for waves 1 to 3. In each of the three problem areas (symptoms, impairment, parental impact), the group that was destined to receive treatment was significantly worse than the nontreatment group. Untreated individuals improved from assessment to assessment; pretreatment
Effectiveness of nonresidential specialty mental health services
387
individuals were deteriorating before treatment. We next examined changes in youths with “need” who never received treatment and the rest of the “need for treatment” sample before they began treatment. We computed interwave change scores for total numbers of symptoms, level of impairment, and level of perceived impact as scores at wave T minus scores at wave T—1. Thus, improvement resulted in negative change scores, whereas deterioration produced positive change scores. For the untreated group we then averaged their change scores across the four waves, because they were to be compared with a single change score (that from pretreatment to first treated wave) in the treated group. Figure 21.2 shows that the pretreatment group were not only more severely disturbed than the nontreatment group, but that they were getting worse on each dimension. In contrast, the untreated group showed improved symptom scores, remaining unchanged on impairment and perceived parental impact. Thus, both symptom level before treatment and change in symptoms before treatment must be controlled in models comparing outcomes for treated and untreated individuals. One obvious corollary of this is that an effectiveness study of this sort requires at least three data points, because with only two it is impossible to control for pretreatment trajectories.
Figure 21.2. “Pretreatment” changes in severity in the untreated and treated groups. Specialty mental health treatment was effective, controlling for initial symptom level and prior symptom change. These analyses were restricted to individuals who were first reported to be in specialty mental health treatment at either wave 2 or wave 3, because they alone had: (a) a previous wave (wave 1 or 2, respectively) of observation, permitting change between pretreatment waves to be computed; and (b) at least one posttreatment wave (waves 3 and 4, or wave 4, respectively). To create a similar comparison time for all cases and the controls, we included only data from the third wave for those first
Annual progress in child psychiatry and child development 2000–2001
388
treated at wave 2 and only from the fourth wave for those first treated at wave 3. Of this treated group, 62% were first treated at wave 2 and 38% first treated at wave 3. We then used a random number generator to select similar proportions of the untreated controls from those waves. Thus, for each subject we had a single outcome observation. The predictor was the number of treatment sessions, and the outcomes of interest were symptoms, impairment, and parental impact at the follow-up waves. We conducted weighted Poisson regression analyses, controlling for level of symptoms, change in symptoms over the year, preceding presentation for treatment, age, and gender.
Figure 21.3. Symptom levels at follow-up by amount of treatment. Figure 21.3 shows adjusted total symptom count means for individuals with different levels of treatment. It can be seen that there was a significant doseresponse effect. Higher levels of treatment were associated with lower levels of symptoms at follow-up. The combination of a significant effect of treatment overall on outcome, plus a clear doseresponse effect of treatment, strongly suggests that treatment reduced symptoms. However, Figure 21.3 shows one striking deviation from this pattern. Those who had only 1 or 2 treatment sessions—who may reasonably be regarded as lacking any reasonable trial of treatment—did worse than those who had no treatment (F1,589=3.8, p=.04) despite an overall effect of treatment. This indicated that those who presented for treatment, but were effectively untreated, were different from those who did not present for treatment. This, in turn, suggests that despite our controls for previous status, we had still failed to control for all the pretreatment differences between treated and untreated individuals, and therefore probably underestimated effects of treatment. The alternative explanation, that those who received only one or two sessions were actually made worse by those sessions, is unlikely to be true. We know that those who presented for treatment
Effectiveness of nonresidential specialty mental health services
389
were more disturbed than those who did not, and that they were getting worse before they presented for treatment. It is not surprising that having presented, and failed to complete any adequate course of treatment, they should have continued to deteriorate. We are also unaware of any other evidence that attendance at one or two sessions at a clinic is harmful to mental health. Nor is there any convincing theoretical or practical reason to suggest that any of the treatments in common use in specialty mental health settings would be harmful in a very low dose, but effective in a higher dose. Another point to note is that real improvement was not clearly demonstrated until an individual had received more than eight sessions. This suggests that very short-term interventions were unlikely to be successful in reducing symptoms. Specialty mental health treatment had no effect on impairment or parental impact over a period of 1 year. Similar analyses with impairment or parental impact as outcomes revealed no significant effects of treatment (Figure 21.3). Greenbaum and colleagues’ (1996) study of treated seriously emotionally disturbed youths also failed to find improvements in impairment. It may be that, as has been found in the treatment of adult depression (Bothwell & Weissman, 1977; Paykel & Weissman, 1974), the time course of improvement of these factors is longer than that for symptoms, so that treatment differences have not yet emerged in our sample.
DISCUSSION Our results provide evidence that the multisector system for providing specialty mental health treatment to children in a rural community was effective in reducing the progress of psychiatric symptoms. Children who were going to receive treatment in the future were already more disturbed than were children who, although significantly symptomatic or psychiatrically impaired, never presented for treatment. They were also on a different trajectory, being less likely to have symptomatic improvement over time than those who never received treatment. At the point at which they entered treatment, treated children had already shown substantial deterioration in symptoms, impairment, and the negative impact of their disorders on their parents. After treatment, this deteriorating trend was either reversed (symptoms) or halted (impairment and parental impact). There was also a significant association between amount of treatment received and degree of improvement in symptoms after treatment. Given these findings, why have earlier studies failed to find effects of treatment? First, most have been small, with less than 50% power to detect a real difference at .05 level in a 2-tailed t-test. Second, they have typically used very global outcome measures (such as “improved,” “unchanged,” or “worsened”) or outcome questionnaire measures that did not include many of the key symptoms that lead to referral for treatment (for example, suicidality, or serious antisocial or aggressive behavior). None included children’s own reports of their mental states, although child self-reports are now regarded as being a key component of the measurement of psychopathology (Rutter & Sandberg, 1985). Nor have differences in disorder trajectory between “treatment” and “control” groups been included in previous studies. It is also possible that recent progress in treatment efficacy research has resulted in the use of more effective treatment over the past decade.
Annual progress in child psychiatry and child development 2000–2001
390
Although only the last of the problems mentioned in the preceding applies to the Fort Bragg Demonstration Project (Bickman et al., 1995), only children who received treatment were studied there, so it was not possible to compare the trajectories of to-betreated and untreated children or make any sort of comparison between untreated and treated children. Limitations The present study has some important limitations. First, it does not address effectiveness in children under the age of 10, because the need to observe equal durations of time during which services could have been received meant that the outcomes were all measured in individuals who were at least 10 years old. Second, these analyses do not include information obtained directly from schools, and more work is needed to determine whether the effects of treatment were seen in functioning at school as rated by school personnel, rather than by child and parent. Third, the present study did not collect information on the type of treatment offered; we do not know exactly what therapy was given or what combinations of treatment were used. Thus, we cannot say whether certain types of outpatient treatment were more effective than others. Fourth, Figure 21.3 shows that the group who had only one or two treatment sessions, and who can therefore reasonably be regarded as being treatment dropouts, did substantially worse symptomatically than those who received no treatment at all. The most likely explanation is that we had failed to control sufficiently for other factors that differentiate treatment seekers from those who receive no treatment. In other words, it is likely that our analyses underestimated the effectiveness of treatment. Implications for Research The use of a large, longitudinal, nonexperimental, community-based study to explore treatment effects naturalistically trades experimental rigor for ecological validity. The usual problem with naturalistic designs is that the signalto-noise ratio is too low for treatment effects to emerge. However, that was not the case here. Given that we have evidence that the usual treatment provided by these specialty services did work, more detailed study is needed of which treatments worked, for which children. Clinical Implications This pattern of results indicates that far from treating the “worried well” or children who would have soon returned to mental health spontaneously, the specialty mental health system selected more severely disturbed individuals who were deterioraring, out of the overall pool of individuals who might be considered to need treatment. In fact, when we attempted to match individuals in treatment with equally symptomatic untreated individuals, we were foiled by the lack of highly disturbed, deteriorating individuals who were not in contact with outpatient services at some point. However, that by no means indicates that all is well with the system. We have presented the first evidence that the mental health care system for children can reduce
Effectiveness of nonresidential specialty mental health services
391
symptoms in the real world. Second, the clear dose-response relationship argues for an effect of the treatment process, not just an effect of getting a child to a treatment setting. The results also suggest that real improvement was not apparent until an individual had received more than eight sessions. And here we see a problem: Many children did not receive eight sessions or more. In other words, the treatment system expended a good deal of its efforts in delivering an apparently inadequate dose of treatment. This indicates that we should concentrate resources on ensuring that the service system is able to deliver episodes of treatment of sufficient length to be efficacious. That, in itself, will be no small task.
ACKNOWLEDGMENT This project was supported by NIMH grant MH-48085, NIDA grant DA11301, and NIMH Center funding (MH-57761).
REFERENCES Achenbach TM, Edelbrock C (1991). Child Behavior Checklist: Youth Self-Report, Center for Children, Youth, and Families, University of Vermont at Burlington. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Angold A.Costello EJ (1995). A test-retest reliability study of child-reported psychiatric symptoms and diagnoses using the Child and Adolescent Psychiatric Assessment (CAPA-C). Psychol Med 25:755–762. Angold A, Costello EJ, Farmer EMZ, Burns BJ, Erlcanli A (1999). Impaired but undiagnosed. J Am Acad Child Adolesc Psychiatry 38:129–137. Angold A, Messer SC, Stangl D, Farmer EMZ, Costello EJ, Burns BJ (1998). Perceived parental burden and service use for child and adolescent psychiatric disorders. Am J Public Health 88:75–80. Angold A, Prendergast M, Cox A, Hanington R., Simenoff E, Rutter M (1995). The Child and Adolescent Psychiatric Assessment (CAPA). Psychol Med 25:739–753. Ascher BH, Farmer EMZ, Burns BJ, Angold A (1996). The Child and Adolescent Services Assessment (CASA): description and psychometrics. J Emotional Behav Disord 4:12–20. Ashcraft CW (1971). The later school adjustment of treated and untreated emotionally handicapped children. J Sch Psychol 9:338–342. Bickman L, Guthrie PR, Foster EM et al. (1995). Eualuating Managed Mental Health Services: The Fort Bragg Experiment. New York: Plenum. Binder DA (1983). On the variances of asymptotically normal estimators from complex surveys. Int Stat Rev 51:279–292. Bothwell S, Weissman MM (1977). Social impaitments four years after an acute depressive disorder episode. Am J Orthopsychiatry 47:231–237. Casey RJ, Berman JS, Castellani S, Pettle WM, Ellinwood E, Alterman R (1985). The
Annual progress in child psychiatry and child development 2000–2001
392
outcome of psychotherapy with children. Psychol Bull 98:388–400. Costello EJ, Angold A, Burns BJ, et al. (1996). The Great Smoky Mountains Study of Youth: goals, designs, methods, and the prevalence of DSM-III-R disorders, Arch Gen Psychiatry 53:1129–1136. DeFries Z, Jenkins S, Williams EC (1964). Treatment of disturbed children in foster care. Am J Orthopsychiatry 34:615–624. Farmer EMZ, Angold A, Burns BJ, Costello EJ (1994). Reliability of self-reported service user testretest consistency of children’s responses to the Child and Adolescent Services Assessment (CASA). J Child Fam Stud 3:307–325. Greenbaum PE, Dedrick RF, Friedman RM et al. (1996). National Adolescent and Child Treatment Study (NACTS): outcomes for children with serious emotional and behavioral disturbance. J Emotional Behav Disord 4:130–146. Hoagwood K, Hibbs E, Brent D, Jensen P (1995). Introduction to the special section:; efficacy and effectiveness in studies of child and adolescent psychiatry, J Consult Clin Psychol 63:683–687. Jensen PS, Hoagwood K., Petri, T (1996). Ourcomes of mental health care for children and adolescents, II: literature review and application of a comprehensive model. J Am Acad Child Adolesc Psychiatry 35:1064–1077. Levitt EE (1957a). A comparison of “remainers” and “defectors” among child clinic patients. J Consult Clin Psychol 21:316. Levitt EE (1957b). The results of psychotherapy with children: an evaluation. J Consult Clin Psychol 21:189–196. Messer SC, Angold A, Costello EJ, Burns BJ (1996). The Child and Adolescent Burden Assessment (CABA): measuring the family impact of emotional and behavioral problems. Int J Methods Psychiatr Res 6:261–284. Paylcel E, Weissman MM (1974). Social adjustment and depression: a longitudinal study. Arch Gen Psychiatry 28:659–663. Richters JE, Arnold LE, Jensen PS, et al. (1995). NIMH Collaborative Mulcisite Multimodal Treatment Study of Children With ADHD, 1: background and rationale. J Am Acad Child Adolesc Psychiatry 34:987–1000. Rutter M, Sandberg S (1985). Epidemiology of child psychiatric disorder: methodological issues and some substantive findings. Child Psychiatry Hum Dev 15:209–233. Seligman MEP (1995). The effectiveness of psychotherapy: the Consumer Report study. Am Psychol 50:965–974. Shepherd M, Oppenheitn AN, Mitchell S (1966). Childhood behaviour disorders and the child-guidance clinic: an epidemiological study. J Child Psychol Psychiatry 7:39–52. US Congress, Office of Technology Assessment (1991). Mental health problems: prevention and services. In: Adolescent Health—Volume II: Background and the Effectiveness of Selected Prevention and Treatment Services. Washington, DC: US Government Printing Office, pp 431–496. Weisz J, Weiss B (1989). Assessing the effects of clinic-based psychotherapy with children and adolescents, J Consult Clin Psychol 57:741–746. Weisz JR, Hawley KM (1998). Finding, evaluating, refining, and applying empirically supported treatments for children and adolescents. J Clin Child Psychol 27:206–216.
Effectiveness of nonresidential specialty mental health services
393
Weisz JR, Huey SJ, Wettsing VR (1998). Psychotherapy outcome research with children and adolescents the state of the art. Adv Clin Child Psychol 20:49–91. Weisz JR, Weiss B, Alicke MD, Klorz ML (1987). Effectiveness of psychotherapy with children and adolescents: a meta-analysis for clinicians. J Consult Clin Psychol 55:542–549. Weisz JR, Weiss B, Donenberg GR (1992). The lab versus the clinic: effects of child and adolescent psychotherapy. Am Psychol 47:1578–1585. Weisz JR, Weiss B, Han SS, Granger DA, Morton T (1995). Effects of psycho therapy with children and adolescents revisited: a meta-analysis of treatment outcome studies. Psychol Bull 117:450–468.
PART V: TREATMENT ISSUES
22 Early Intervention Programs for Children with Autism: Conceptual Frameworks for Implementation Heather Whiteford Erba
Four diverse early intervention programs for children with autism— discrete trial training, LEAP, floor time, and TEACCH—are described. For each pro gram, the concepts of learning, development, and autism are summarized, intervention procedures are outlined, and connections between theory and practice are illustrated. Research outcomes for each of the four programs are discussed.
INTRODUCTION Autism is the best-documented disorder in child psychiatry (Peeters, 1997). However, questions faced by Kanner (1943), when he recognized infantile autism as a new syndrome, still plague researchers today. Cohen and Volkmar (1997) aptly address the complexity inherent in this disorder: “Virtually every type of theory relating to child development, cognitive, social, behavioral, affective, and neurobiological, has been applied to understanding the enigmatic impairments and competencies of autistic individuals” (p. xv). Currently, autism is understood as an organic developmental disorder affecting a child’s ability to interact with and experience his or her world (Siegel, 1996); although prognoses are diverse, it is viewed as a chronic disability that results in lifelong impairment for most individuals (Rogers, in press). During the last decade, great gains have been made in understanding the structural and behavioral nature of this disorder as research has shown positive outcomes in young children receiving early intervention services (Rogers, in press). Findings on early intervention suggest that the functional potential of children with autism can be increased following intensive programs (Rogers, 1996). Recognizing this, the National Institutes of Health (NIH) has suggested a correlation with the plasticity of the brain at this early age (Bristol et al., 1996). Results of a retrospective study (Fenske, Zalenski, Krantz, & McClannahan, 1985) supported the hypothesis that children with autism have significantly better outcomes if the intervention program begins prior to age 5 than if it does so after age 5. Four tenets supporting the efficacy of early intervention for young children with autism
Annual progress in child psychiatry and child development 2000–2001
396
and problem behavior have been proposed by Dunlap and Fox (1996). First, early development corresponds directly with the acquisition of communication skills, and early social-communication experiences are the basis for future language development and social interactions. Because of the impaired abilities in communication and social relatedness apparent in autism, successful interventions must address these issues, taking advantage of the developmental activity of these early years. Development of communication abilities has been shown to decrease children’s problem behavior and increase their ability to interact successfully with peers (Fox, Dunlap, & Philbrick, 1997). Second, because it is easier to functionally analyze the meaning of children’s behavior at an early age, individual programs that better address the unique needs of the child can be designed in early intervention. Third, greater collaborative success has been observed (Dunlap & Fox, 1996) with families of young children with disabilities than with those of older children or adults. Because the family system and developmental environment directly affect a child’s ultimate success, it is imperative to involve the family in every aspect of intervention. To optimize this relationship, intervention should take place as early as possible. Finally, autism is often associated with challenging types of behavior that may place child, family, and peers at risk. Typically, it is easier to develop behavioral programs while the child is physically small; also, reducing such aggressive and self-injurious behavior as head banging and biting must be addressed as soon as possible. Early intervention services are mandated for all children with disabilities (PL 99–457 and PL 102–119), and are empirically supported and justified for children with autism. Several diverse programs have published efficacy reports supporting the positive outcomes of their programs (Dawson & Osterling, 1997; Harris & Handleman, 1994). According to NIH (Bristol et al., 1996) “Treatments that are dramatically effective for one person with autism may be ineffective or even contraindicated for others” (p. 5). Although several comprehensive early intervention programs have common components, their underlying theories often conflict. Any program’s theoretical framework will influence its strategies and evaluation methods (Meisles, 1985). Dawson and Osterling (1997) have suggested that the underlying philosophies of programs are not critical to positive outcomes as long as fundamental programming components are evident. However, families, doctors, and service providers need information on a program’s philosophy in order to make informed decisions regarding intervention strategies. The intent of the present article is to illustrate the structural framework apparent in each of four diverse early intervention models. These are discretetrial training (DTT), based on the work of Lovaas and colleagues (Lovaas, 1987; McEachin, Smith, & Lovaas, 1993); Learning Experiences…An Al-ternative Program for Preschoolers and Parents (LEAP) (Strain & Cordisco, 1994); floor time (Greenspan & Wieder, 1998); and Treatment and Education of Autistic and Related Communication Handicapped Children (TEACCH) (Mesibov, 1996). In describing each, theories of learning, development, and the nature of autism as conceptualized by the model are summarized. Next, issues of reinforcement control (basing behavioral change on control of consequences) versus stimulus control (manipulating environment or antecedents to control behavior) within the program are discussed. Then, taking the key theoretical components into consideration, the model’s intervention technique is described. Finally, the link between
Early intervention programs for children with autism
397
the program’s core theory and outcome data for children is examined.
EARLY INTERVENTION PROGRAMS The four programs discussed here are comprehensive and attempt to improve the overall functioning abilities of persons with autism (Rogers, in press), though they represent very different theoretical frameworks. Efficacy data published in peer-reviewed articles are cited (due to space considerations, other programs that meet the same standards have not been included). They are nationally recognized, and are increasingly desired by families of children with autism throughout the country (National Early Childhood, 1997). DTT Program The work of Lovaas and colleagues at the Young Autism Project at the University of California, Los Angeles (UCLA) has played an important, albeit controversial, role in the current understanding of learning and optimal outcomes for young children with autism. A “Lovaas Program” (referred to here as a DTT program) is based on an operantconditioning behavioral model that relies predominantly on intensive discrete-trial sessions. An understanding of applied behavioral analysis (ABA)—implementation and evaluation of the wide range of principles and techniques that constitute behavioral learning theory (Wolery, Bailey, & Sugai, 1988)—is essential to comprehending the nature of Lovaas’ behavioral theory of autism (Lovaas & Smith, 1989). A conceptual overview of behavioral theory and the ABA process therefore follows. Behavioral learning theory is based on three core assumptions: (a) behavior is conceptualized within a three-term contingency that includes antecedents, behavior, and consequences; (b) antecedent stimuli and prior experience of consequences will affect behavioral reactions; and (c) effective teaching incorporates control of antecedents and consequences (Wolery et al., 1988). Skinner’s (1968) operant conditioning model, on which the DTT program is based, focuses on the use of positive reinforcement as the key agent of behavioral change. Operant conditioning principles, such as shaping, chaining, discrimination training, and contingency management, are used within a discrete-trial paradigm (Smith & Lovaas, 1998). Its theory is based on reinforcement control as the basis for behavior change; that is, behavior that is positively reinforced will continue, whereas behavior that is ignored or punished (by time-out, aversives, a verbal “no”) will stop (Lovaas & Smith, 1989). The primary technique used throughout the Lovaas model is the discrete-trial session, which consists of four parts: (a) the trainer’s presentation of stimuli to which a child responds; (b) the child’s response; (c) the consequence; and (d) a short pause prior to the next command (Anderson, Taras, & O’Malley-Cannon, 1996). To avoid the misconceptions and overgeneralizations prevalent in the field, it is important to note that a DTT program is not synonymous with ABA. The latter is a dynamic process based on more than 50 years of wide-ranging scientific research (Green, 1996); its basic goal is effective, individualized instruction based on principles of behavior theory (Wolery et al., 1988). DTT is a single method associated with this process. Several specific focal treatments (interventions targeting a single type of
Annual progress in child psychiatry and child development 2000–2001
398
behavior) have successfully used a range of ABA techniques with children with autism (Rogers, in press); these include pivotal response training, a natural language paradigm, and systematic generalization techniques (Koegel & Koegel, 1995). Conversely, a DTT program, as defined by Lovaas (1981), is a comprehensive behavioral program seeking to improve the overall outcomes of children with autism. Theoretical Foundation. DTT applies the theories and principles of Lovaas’s behavioral theory of autism (Lovaas & Smith, 1989) specifically to children with autism. Behavioral principles of learning were first applied to children with this disorder by Ferster (1961), who had worked with Skinner during the development of the operantconditioning model (Ferster & Skinner, 1957). Accepting the prevalent psychogenic theory of the time, Ferster’s behavioral hypothesis suggested that children with autism were not receiving positive reinforcement in the form of social praise or parental attention. Because of their consequent inability to equate attention or social praise with positive reinforcement, the children were not learning naturally in the environment. Ferster’s work (1961) illustrated that operant-conditioning techniques, using primary reinforcements (e.g., food), had positive affects and could foster learning in children with autism. The current organic understanding of autism empirically refutes the notion that parental behavior is instrumental in the disorder’s etiology; nevertheless, the ability of these children to respond to behavioral techniques opened the doors to several successful behaviorally based intervention techniques. Ferster’s (1961) work was developed by Lovaas and Smith (1989) into their current behavioral theory of autism. (Caution should be observed in applying this theory to other investigators using ABA techniques with children with autism, because this theory is the framework for a specific DTT program.) The DTT theory of autism has four tenets: (a) behavioral principles can be successfully applied to children with autism; (b) multiple unique behavioral difficulties are evident in children with autism; (c) when placed in special environments, children with autism are able to emulate typical learning patterns; and (d) a mismatch is apparent between autistic children’s nervous systems and the environment (Lovaas & Smith, 1989). This view rejects the central-deficit theory of autism, which suggests strengths, deficits, and a specific neurological disorder common to all children with autism. Conversely, it proposes a multiplicity of unique types of behavior in the children, citing the wide range of abilities and differences to be found in individuals, and the lack of behavior specific to the autistic population as a whole. Procedures. The Lovaas model follows a 2- to 3-year program that was tightly monitored by the Young Autism Project at UCLA. The first phase of intervention includes 40 hours per week of one-on-one discrete trial training, administered by a team of trained university students and the children’s parents over a 1- to 2-year period. This phase teaches the children skills that their typically developing peers in the nonclinical population learn in their natural environment. Based on the theory that a mismatch exists between autistic nervous systems and the environment, and that children with autism can emulate natural learning paradigms in special environments, the first phase is implemented in an isolated one-on-one setting. According to Lovaas and Smith (1989), “…at the beginning of treatment, the children may be regarded as having close to a tabula rasa” (p. 23). Activities such as teaching nonverbal imitation skills and match-to-sample tasks are used to optimize children’s
Early intervention programs for children with autism
399
outcomes. Specific individual traits (i.e., self-stimulatory and/or selfinjurious behavior) are also targeted at this time. The second phase focuses on expressive and receptive language skills, abstract play, and social behavior, both within the context of one-on-one discrete-trial training, and generalized to playgroup and/or supported preschool experiences (Dawson & Osterling, 1997; Lovaas, 1981). This phase reflects the notion that children with autism will successfully respond to basic behavioral principles. The program strongly urges that the children be placed in inclusive settings to “normalize” their social abilities (Lovaas, 1981). The placement may involve the support of an inclusion facilitator but, ideally, the children will be interacting independently with their peers in a typical classroom. Often, academic skills are taught in the one-on-one discrete-trial format so that the children can focus on social and communication skills in the group setting. The intervention model reflects the key theoretical components of the model. Outcomes. Lovaas and his colleagues have published the strongest empirical support to date for a specific early intervention program (Rogers, in press). Initially, a group of 19 young children with autism received 40 hours per week of DTT over a 2-year period. When compared to two control groups, one receiving 10 hours per week of DTT and the other receiving typical community support, the sample treatment group was reported to enjoy significantly higher rates of success; specifically, 47% had reportedly recovered and was functioning well in inclusive first-grade classrooms (Lovaas, 1987). The results of a 1993 follow-up illustrated the stability of the initial outcomes. At an average age of 11.5 years, eight out of nine members of the original treatment sample were reported as indistinguishable from their same-age peers (McEachin, Smith, & Lovaas, 1993). Currently, eight sites are replicating the treatment. Unfortunately, outcome data are not yet available (Smith & Lovaas, 1998). The success of intensive DTT has arguably provided children and parents with unexpected outcomes. At the same time, it has instigated a methodological controversy. Claims of “recovery” and “indistinguishable” functioning abilities have been questioned by several professionals in the field (Kazdin, 1993; Schopler, Short, & Mesibov, 1989). Also, methodological issues of sampling bias and lack of random assignment have tainted the findings. Additional research and independent replications of the program are needed to clarify the effects of DTT programs on children with autism. LEAP Intervention The LEAP intervention model combines developmentally appropriate practice (DAP) and ABA techniques in an inclusive program (Strain & Cordisco, 1994) in which several diverse learning theories have been blended to create the conceptual framework. Although it recognizes the range of strengths and needs apparent in children with autism, the primary focus of the program’s underlying theory and practical implementation is the children’s social development. In this application, therefore, the theory of autism revolves around a central social deficit. With its diverse theoretical influences, the LEAP model uses both reinforcement and stimulus-control teaching techniques. Its guiding principles (Strain & Cordisco, 1994) are: (a) all children benefit from integrated environments; (b) autistic children benefit when intervention is consistent across school, home, and
Annual progress in child psychiatry and child development 2000–2001
400
community; (c) they make better gains when parents and teachers work together; (d) they can learn from typical same-age peers; (e) intervention should be planned, systematic, and individualized; and (f) children with and without disabilities benefit from activities that reflect DAP. LEAP’S conceptual framework is based on theories of behavior, key principles of DAP, and inclusion. Theoretical Foundation. The LEAP model is directly influenced by behavioral learning theories. The behavior prompting, fading, and reinforcement techniques embraced by Lovaas and colleagues are also implemented in this model to foster independent play skills, social interactions, and targeted behavior (Dawson & Osterling, 1997). The LEAP model, as is illustrated by its dedication to inclusive practices, also embraces naturalistic ABA techniques such as milieu language techniques (Warren & Kaiser, 1988), modeling, and planned generalization strategies (Strain & Cordisco, 1994). Fundamentally, LEAP is influenced by a belief in social learning theory (Bandura, 1969), Bandura’s model of behavioral learning theory suggested that learning can take place through observations of others’ behavior and its consequences. The LEAP model recognizes that children with autism need support and assistance in order to learn vicariously, through others, but is dedicated to the notion that all children’s can learn and benefit from this technique. A range of ABA methods is implemented to expose children with autism to developmentally appropriate preschool activities. DAP is a term developed by the National Association for the Education of Young Children (NAEYC) and was derived from their commitment to promoting high quality programs for all children and families (Bredekamp, 1997). LEAP strives to adhere to DAP guidelines, which have three components: (a) knowledge of child development and learning; (b) knowledge of the individual strengths, interests, and needs of each child; and (c) knowledge of each child’s cultural and social background. Consequently, they are both individually and age-appropriate (Bredekamp, 1997). The LEAP model integrates key developmental components recognized by NAEYC, while implementing a range of ABA techniques typically associated with early childhood special education in an inclusive setting. Debate has revolved around the notion of, inclusion since the 1970s. Although an indepth discussion is beyond the scope of this article, a brief comment on its theory and practice is necessary to understand LEAP’S dedication to inclusive services. The Division of Early Childhood (1993) stated, “Inclusion, as a value, supports the right of all children, regardless of their diverse abilities, to participate actively in natural settings within their communities.” Guralnick (1990) saw inclusion as reflecting the core values of equity, full participation, and acceptance, all of which have since become apparent in the field of early childhood education. The arguments postulated in support of inclusive services have been categorized by Hanline and Galant (1993) as legal-legislative, socialethical, and psychological-educational. Research findings have supported the use of inclusive settings in promoting social integration (Guralnick, 1990) and general developmental outcomes (Bricker, Bruder, & Bailey, 1982; Jenkins, Speltz, & Odom, 1985). A common argument against inclusion is that typically developing children will model undesirable behavior by peers with special needs. The results of research on preschool LEAP have refuted this argument, suggesting that typically developing peers in the inclusive program displayed less deviant behavior than did matched peers identified
Early intervention programs for children with autism
401
as classroom “stars” (Strain, Hoyson, & Jamieson, 1985). Research and practice have also revealed the challenges faced by inclusive programs. To be suc-cessful, they must go beyond simple physical integration (Odom & Brown, 1993). Bricker (1995) suggested positive attitudes, resources, and activity-based curricula as key components of positive inclusion experiences for children, families, and classroom staff. LEAP’S guiding principles reflect these components, and illustrate their theoretical commitment to inclusive practices. Procedures. LEAP’S focus on behavioral, DAP, and inclusive ideals, are reflected in its intervention techniques. The program consists of an integrated preschool classroom, behavioral skill training for parents, and outreach training services, components that reflect the program’s diverse theoretical framework. The integrated classroom comprises ten typical and six autistic children. The program operates for 3 hours daily, year round. Parents of all children may participate in classes designed to teach behavior management and strategies for teaching new skills. A key LEAP component, using behavioral techniques in an inclusive group setting, is peer-based intervention, in which typically developing peers act as indirect mediators of behavior change, behavior models, and direct agents of training (Strain et al., 1985). Group-oriented (interdependent reinforcement) contingencies are used to reinforce individual students’ behavioral goals within the context of the group setting. All the children have individual goals but must perform to a minimum standard before these will be positively reinforced. This behavioral strategy focuses on independent goals, while fostering group participation and interaction. Peer-imitation training is used to optimize opportunities for autistic children to copy typically developing peers in daily sessions of approximately 30 minutes. Activity-based intervention (Bricker & Woods-Cripe, 1992) and systematic planning are also used to ensure ample opportunity for them to observe and imitate peer behavior in natural settings. This intervention technique is directly tied to Bandura’s (1969) social learning theory. Typical peers may be chosen as trainers and taught how to engage and work with children with autism, i.e., how to persist in play and cope with unsuccessful play attempts without disappointment or confusion. Positive reinforcement techniques are used with both peer trainers and the target children. Instruction for children with autism is designed to resemble that for typical peers, focuses on functional skills, incorporates planned activities into those selected by the children, is individualized to fit each child, and should promote generalization, resulting in acquisition of the desired skill by the child (Strain & Cordisco, 1994). Outcomes. The LEAP program is one of the few intervention models to publish longitudinal outcomes for their participants. As with the Lovaas model, LEAP’S success is based on children’s social abilities and is typically measured by kindergarten inclusion. Approximately 50% of students attending the LEAP program are reported as successfully attending “regular education classes” (Dawson & Osterling, 1997). Strain, Kohler, and Goldstein (1996) reported a significant reduction of autistic symptoms following 24 months of treatment, and that 24 of 51 LEAP children were subsequently included in a public school setting.
Annual progress in child psychiatry and child development 2000–2001
402
Floor Time Theoretical Foundation. The floor time approach is based on a developmental interactive theory, according to which most cognitive skills developed in the first 4 or 5 years of life are based on emotions and relationships (Greenspan & Wieder, 1997a). Thus, affect and interactive relationships are the primary components in the theory and practice of this model. Greenspan and his colleagues have developed an integrated developmental approach to intervention with children who have severe difficulties in relating and communicating, and a systematic interactive intervention technique known as floor time. Among the influences on the conceptual framework of this program are the importance of relationships, six specific social milestones, and a hypothetical theory of autism. Greenspan and Wieder (1997a) have suggested that: “The child’s interactions in relationships and family patterns are the primary vehicle for mobilizing development and growth” (p. 5). Relationships provide the necessary support for children to develop key abilities. First, through reinforcing relationships, children experience security, warmth, pleasure, and physical safety that allow them to learn important self-regulatory skills and the ability to be aware, alert, and attentive to their environment. Early relationships provide the basis for two-way communication through nonverbal and verbal exchanges in which subtle affective cues convey whether a behavior is appropriate or inappropriate. Symbolic and representational abilities begin to develop within the context of reinforcing relationships. Children must use emotional experiences to develop true representational and abstract thinking abilities (Greenspan & Wieder, 1997a). A wide range of family patterns can foster positive relationships, as can individual child characteristics. The goal of the floor time interactive model is to establish secure relationships in order to foster development. According to the model, all children must master six “functional emotional skills” that “underlie all our intelligence and interactions with the world” (Greenspan & Wieder, 1998, p. 3). These skills are the ability to: (a) self-calm and process environmental information; (b) engage in relationships; (c) engage in two-way communication; (d) create complex gestures, and connect a series of actions into an elaborate and deliberate problem-solving sequence; (e) create ideas; and (f) build bridges between ideas so that they become reality-based and logical (Greenspan & Wieder, 1998). These are milestones in a sequential developmental pattern, and form the basis for critical cognitive, socialemotional, language, and motor skills development. This milestone theory should not be confused with cognitive-developmental theories based on the works of Piaget (1952). Greenspan has suggested that typically developing children who enjoy nurturing relationships will master these milestones within the context of their natural environment, but that children with developmental challenges may need support or direct intervention in order to achieve such mastery. The floor time theory is organic in nature and recognizes symptoms of children with autism as secondary manifestations of underlying sensory dysfunction. Processing difficulties may include variations in auditory understanding, motor planning, and sensory modulation. This hypothesis proposes that the primary neurophysiological dysfunction may lie in the connection between affect and the sequencing of motor
Early intervention programs for children with autism
403
patterns and verbal symbols (Greenspan & Wieder, 1997b). Thus, the primary goal of intervention is to help children work around processing difficulties so as to reestablish affective contact with primary caretakers and begin to master the six relationship-based milestones (Greenspan & Wieder, 1997b). Procedures. An integrated developmental approach to intervention uses a sequential technique focusing on individual child and family characteristics. The model is conceptualized as an integrated developmental intervention pyramid. The levels, which must be addressed in turn, range from base to apex and include: (a) basic safety, protection, and security; (b) development of ongoing trusting relationships; (c) implementing relationships geared to the differences in each child’s individual processing; (d) matching relationship-based interactions to the child’s developmental level; and (e) specific interventions (Greenspan & Wieder, 1997a). Greenspan and Wieder have recognized and validated a range of techniques that can be used as specific interventions, including the floor time approach, inclusive peer-based models (Strain & Cordisco, 1994), intensive discrete-trial approaches (Lovaas, 1981), and various focal treatments implemented by speech pathologists and occupational and physical therapists. Although Greenspan advocates and supports floor time as a developmental interactive approach, the individual needs of the child and family drive actual technique selection. This model of intervention development emphasizes that building affective relationships while dealing with children’s sensory processing issues must be considered in the context of every specific intervention technique. For example, if families choose to implement an intensive DTT program, it is recommended that that floor time methods be ultimately woven into the intervention in order to generalize skills and build relationships. Unlike therapist-directed DTT techniques or the systematically planned LEAP approach, floor time uses a child-directed play period. Following formal and informal assessments of children’s unique processing abilities, strengths, and developmental needs, therapists design a developmentally based interactive plan. The children direct play throughout the session, and the therapists are challenged to follow their lead and respect their interests, while creating situations through the play that will address the children’s emotional needs. The floor time model recommends that parents be the first and primary play partners. The technique may then be taught to speech and language pathologists, occupational therapists, and developmental therapists, For example, if a child’s problem-solving abilities are limited, a therapist may set up several problems during the course of a session (e.g., a dinosaur gets stuck in a hole, a train track is missing). Through this playful relating and engagement, the theory hypothesizes that children will be able to compensate for their processing problems and be able to master the developmental milestones. The developmental approach to therapy begins with parent-child floor-time sessions, generalizes to specialists (including speech, occupational, physical, and educational therapists) who use floor-time techniques to target specific processing deficits, and continues with families’ constant efforts to relate to their child at higher developmental levels. Floor time targets four specific goals that correspond to the six developmental milestones: (a) to encourage attention and intimacy (milestones 1 and 2); (b) to build twoway communication (milestones 3 and 4); (c) to support expression of feelings and emotions, and the use of ideas (milestone 5); and (d) to support logical thinking
Annual progress in child psychiatry and child development 2000–2001
404
(milestone 6; Greenspan & Wieder, 1997a). The floor-time approach should not be confused with play therapy or with psychotherapy that uses play as a medium to build affect but does not address underlying processing abilities. The floor-time model uses each child’s emerging developmental skills as guides to cognitive and developmental growth through affective relationships (Greenspan & Wieder, 1997b). As do the other models discussed here, this technique recognizes the need for intensive early intervention, suggesting eight or more 20- to 30-minute floor-time sessions per day (Greenspan & Wieder, 1997a). The floor-time intervention technique corresponds to the program’s underlying conceptual framework. Outcomes. Greenspan and Wieder (1991b) have illustrated the success of a floor-time approach for children with autism in a chart-review publication. The review compared the charts of 200 children with autistic spectrum disorders to those of 53 comparison children with autism receiving community intervention support. Outcomes were assessed using the Functional Emotional Assessment Scale (Greenspan & Wieder, 1998), and categorized as good to outstanding, medium, or having ongoing difficulties. All children were assessed after 2 or more years of a floor-time intervention plan. Of the 200 in the experimental group, 58% fell into the good-to-outstanding category, showing spontaneous symbolic abilities that related to intent and affect. These children were also able to engage in purposeful, organized, and symbolic problem-solving sequences, and had developed multiple relationships. On the Childhood Autism Rating Scale (Schopler, Reichler, & Renner, 1985), rates for this group fell outside the autistic range. The 24% of participants who fell into the medium category showed good mastery of the first three or four milestones, but were just beginning to develop symbolic abilities. The 17% who fell into the third category revealed ongoing difficulties in relationships and in understanding symbolic representations (Greenspan & Wieder, 1997b). In Table 22.1, outcomes for children receiving floor time are compared with those for children receiving a range of community-based early intervention options, including DTT, public preschool, and focal treatments. Although Greenspan and Wieder’s chart review is not considered an empirical study, the results support the intervention and warrant sound scientific inquiry. TEACCH Program Division TEACCH is a statewide program serving individuals with autism in North Carolina, and an internationally recognized service delivery system (Mesibov, 1996). In contrast to the other three programs discussed here, it provides a lifelong continuum of services for individuals, families, and service providers. Services include assessment/diagnosis, treatment, consultation, community collaboration, supported employment and living, and a range of services specific to family needs. TEACCH therapists are considered generalists—knowledgeable in multiple fields, including (but not limited to) speech pathology, social work, early intervention, special education and psychology (Lord, Bristol, & Schopler, 1993). The conceptual framework of the TEACCH model is based on behavioral, developmental, and ecological theoretical perspectives and tied directly to an organic theory of autism. Theoretical Foundation. Mesibov & Shea (1998) stated the TEACCH theory of autism thus: “Because the organically-based problems that define autism are not reversible, we
Early intervention programs for children with autism
405
do not take ‘being normal’ as the goal of our educational and therapeutic efforts. Rather, the long-term goal of the TEACCH program is for the student with autism to fit as well as possible into our society as an adult.” The TEACCH theory of autism, a central-deficit theory, suggests that individuals with the disorder share certain learning problems and strengths (Lovaas & Smith, 1989). Although young children with autism show fewer skills in imitation, initiation, and pretend play than do typically developing peers (Rogers & Pennington, 1991), TEACCH recognizes that all preschool children need, “warmth and structure offered by teachers with clear goals and the flexibility to instill a joy of learning in their students” (Lord et al., 1993, p. 206).
TABLE 22.1 Floor-Time Outcomes
Progress Good to outstanding Medium Ongoing difficulties Greenspan and Wieder, 1997b.
Floor Time
Traditional 58% 24% 17%
2% 40% 58%
Five key principles form the basis for the TEACCH model: (a) use of strengths and interests to build a bridge between the two cultures; (b) careful, ongoing assessment that optimizes opportunities for independence and success; (c) provision of environmental structure that will assist individuals with autism to understand meaning; (d) reframing of noncompliance as the inability of individuals to understand what is expected of them; and (e) involvement of parents as key collaborators on an interdisciplinary team (Mesibov & Shea, 1988). These components are evident in the TEACCH commitment to using behavioral techniques within a developmental framework, while keeping a focus on community and families. The cognitive-developmental learning approach is associated with Piaget’s (1952) stage theory, which proposed four sequential developmental stages through which children move as they acquire biological maturation and environmental experience. This approach to intervention is commonly used for children with autism (Olley & Gutentag, in press), and is reflected in TEACCH’s organic theory of autism (which identifies basic deficits in cognitive processing), as well as in its commitment to individual planning through ongoing assessment. A developmental approach allows TEACCH therapists to recognize both differences between children and differences in each child’s skill attainment levels (Lord et al., 1993). Specific to preschool children, developmental patterns determine such practical issues as when to begin toilet training, and the individualization of varying levels of structure. Within the developmental framework, behavioral techniques such as task analysis, positive reinforcement, and extinction are used to foster independence and success in the environment. On the basis of the central deficit theory, TEACCH suggests that children with autism often require increased structure and task-analyzed goals in order to learn. Direct operant conditioning techniques may be used initially with young children to reduce stereotypic or self-injurious behavior and increase attention. However, TEACCH does not rely on
Annual progress in child psychiatry and child development 2000–2001
406
these techniques to teach the range of social and communication skills targeted in the preschool years. According to Lord and colleagues (1993), “the emphasis in the preschool class…is on learning to be a student and developing appropriate social and communicative behaviors” (p. 212). The child’s environment is designed in terms of physical structure, posted schedules explaining the daily classroom routine, individual schedules clarify each student’s specific activities, a visual independent work system, and clearly organized instructional materials to foster environmental understanding and independence, and to decrease problem behavior. When children’s abilities are blocked by the organic nature of the disorder, the environment is modified to accommodate the identified deficit. The structured teaching aspect of the intervention program employs stimulus-control behavioral techniques in a developmentally based setting, and behavioral principles are also used during the ongoing formal and informal assessment process. When targeting behavior, teachers, families, and others involved in the child’s life consider specific antecedents and consequences of the behavior. To foster the child’s success, behavioral plans are designed using such techniques as visual structure, specific structured teaching activities, or reinforcement procedures. Family involvement has been a key theoretical component of the TEACCH framework from the program’s inception. TEACCH was developed in reaction to the psychogenic theory of autism prevalent in the 1960s, which placed the blame of the disorder on children’s parents. In 1964, Schopler and Richer received a National Institute of Mental Health grant, “Parents as Cotherapists,” that allowed them to pursue research and treatment based on an organic understanding of autism at the University of North Carolina at Chapel Hill. Parents were viewed as key participants in every aspect of the program (Mesibov, 1996). This idea has had lasting and positive implications for the field, and exemplifies the foundation of the TEACCH philosophy. The family-focus has now broadened to include community needs, thus adopting an ecological perspective. “It is the goal of the TEACCH program to build on the strengths of each child, family, and community to help the child achieve particular goals and to help the family and community cope with the challenge of living with a young child with autism” (Lord et al., 1993, p. 219). Parent-teacher collaboration is considered the best way to generalize learning, and develop a positive relationship between the family and TEACCH (parents of children in the preschool program often volunteer in the classroom once or twice a week). Procedures. Unlike the detailed intervention techniques prescribed by DTT programs and LEAP, the TEACCH model calls on a wide range of techniques and services to meet the individual needs of children and families. “There is a strong commitment to drawing from a variety of alternatives to find the right place and the right approach for each child in his or her family at a particular time” (Lord et al., 1993, p. 208). Thus, the key components of the model parallel the conceptual framework: a developmental approach, the importance of family and community, and a direct relationship between assessment and intervention. In addition, behavioral principles are employed throughout the model, particularly during the assessment process and development of behavioral plans. Unlike the other three programs, TEACCH does not consider children’s kindergarten or school placement the measure of success. Issues of inclusion are considered on a case-by-case basis. The ultimate goal is to foster independence and understanding, while providing
Early intervention programs for children with autism
407
children with the tools they will eventually need to interact successfully as adults in the environment. Outcomes. Although research findings have supported specific components of the program’s intervention strategies for individuals of all ages, no empiri-cally supported efficacy study of the comprehensive preschool program has yet been published (Rogers, in press). Lord and Schopler (1989) demonstrated that autistic children first assessed between 3 and 4 years of age had improved IQ scores at 7 and 9 years, regardless of the intensity of the intervention. Ozonoff and Cathcart (1998) compared children receiving structured teaching home programs with matched peers relying predominantly on DTT. Results showed children involved in the less intensive TEACCH model with outcomes three to four times greater than the control group on all outcome measures. Mirroring the need for additional research apparent throughout the field, more data are needed for better understanding of the effects of the TEACCH model on young children with autism.
CONCLUSION The theoretical framework of each of these four programs guides its intervention techniques. Table 22.2 illustrates the key components of each theory and the corresponding intervention techniques. In three of the four programs, behavioral learning theory is an aspect of the overall framework. However, the ABA techniques implemented are diverse, ranging from exclusively reinforcement control to primarily stimulus control. Each program’s unique theory of autism directly modifies the focus of the intervention. A DTT program focuses on deficits; therefore, its intervention technique targets individual deficits and attempts to change behavior model (Lovaas & Smith, 1989). The LEAP model embraces a social learning model of behavioral theory, with an emphasis on inclusion and developmentally appropriate practice. The program’s inclusive intervention technique focuses on children’s ability to learn from each other, and the ability of children with autism (Strain & Cardisco, 1994). Although LEAP and floor time both emphasize the social component of learning, the underlying theories are very different. Floor time is based on a developmental interactive theory that stresses the importance of relationships and affect as bases for developmental progress. Floor time’s milestone approach considers the primary deficit in children with autism to be connected to sensory processing issues that can theoretically be alleviated through intensive individualized early intervention, with a focus on relationships (Greenspan & Wieder, 1998). Finally, TEACCH model is an eclectic approach that merges behavioral and cognitivedevelopment theories and emphasizes individualization and family collaboration. It is the only one of the four models discussed here that is not limited to an early intervention program. The TEACCH theory of autism recognizes primary deficits associated with meaning in individuals with autism. Therefore the goal of the individualized intervention techniques, including but not limited to structured teaching, is to provide supports based on those strengths that foster understanding in the environment (Lord et al., 1993).
Annual progress in child psychiatry and child development 2000–2001
408
TABLE 22.2 Programs for Children with Autism: Theories and Intervention Strategies
Program Underlying Theories
Theory of Autism
Recommended Stimulus Vs Intervention Reinforcement Strategy Control DTT Behavioral learning Multiple Intensive discrete- Reinforcement (operant conditioning) deficit trial training (40 control hours per week) LEAP Behavioral learning Central Inclusive, Both (social learning) deficit: social preschool, DAP abilities curriculum, ABA techniques, peer intervention training FloorDevelopmental Central Individualized, NA Time interactive deficit: based on family processing needs. Floor time. abilities TEACCH Behavioral learning, Central Individualized, Both, primarily cognitivedevelopment deficit: based on family stimulus meaning from needs. Structured control environment teaching. The differences between these programs are apparent at both theoretical and practical levels. Table 22.3 outlines components common to at least two of the programs—family involvement and a commitment to individual programming, for instance, are common to all four programs—though the models vary on how these constructs are actually implemented. Programs may develop interventions with a deficit- or strength-based approach. A DTT program is individualized on the basis of the child’s unique deficits, but the intervention itself follows a specific standardized curriculum; the child’s profile will dictate the program’s starting point, but all children must master similar sets of drills. In the Lovaas programs (Lovaas, 1987; McEachin et al., 1993) and at their current replication sites, parents are asked to provide 10 of the 40 hours of therapy each week and are trained to implement this specific protocol. In the floortime model, too, parents are trained as therapists, and programming is based on children’s deficits. Both floor time and DTT are implemented primarily in the home. Conversely, the TEACCH early intervention and LEAP programs provide a center-based model, with consultative support and carry-over in the home. Parents are actively involved in program development, but are not asked or expected to provide one-on-one intervention with their children. The TEACCH model is strength-based, whereas LEAP primarily follows a deficit based
Early intervention programs for children with autism
409
TABLE 22.3 Programs for Children with Autism: Commonality Analysis
Program ABA Family Individualized Intensity Setting Techniques Involvement Programming DTT Yes Yes Yes 40 hours pw Home May be generalized to inclusive pre school or kindergarten LEAP Yes Yes Yes 3 hours pd. Center. Parent 5 days pw, training provided Inclusive, to support round, preschool consistency in home Floor No Yes Yes B 20–30 Home Time minute sessions pd TEACCH Yes Yes Yes 5 hours pd. Center. 5 days pw, Consultation year-round, provided to Preschool support consistency in home model. All four programs recognized and address the benefits of early intensive programming; however, the recommended amount of intervention ranges from 15 to 40 hours per week. Intensity, then, is a complex issue in need of continued research. Although conflict, primarily based on methodology and design, surrounds the efficacy studies for these interventions, each program has reported successful child outcomes. Research is needed to explore the relationship between the programs’ conceptual frameworks as they relate to specific intervention techniques and child outcomes. In addition, determining how child and family characteristics correlate with intervention methods would increase the likelihood of meeting individualized goals. Current research findings do not give this information, although it is much needed by service providers. Rogers (in press) has pointed out that it is the responsibility of the professionals in the field to be aware of the multiple intervention options for children with autism, provide parents with current and accurate information regarding various validated and unvalidated approaches, and try to provide validated intervention options for all individuals with autism. The task is daunting and complex. Understanding the conceptual frameworks of various programs and the con-nection between their theory and practice is a step toward greater comprehension of intervention options.
Annual progress in child psychiatry and child development 2000–2001
410
REFERENCES Anderson, S.R., Taras, M., & O’Malley-Cannon, B. (1996). Teaching new skills to young children with autism. In C.Mautice, G.Green, & S.C.Luce (Eds.), Behavioral intervention for young children with autism; A manual for parents and professionals (pp. 181–193). Austin, TX: Pro-Ed. Bandura, A. (1969). Principles of behavior modification. New York: Holt, Rinchart, & Winston. Bredekamp, S. (Ed.). (1997). Developmentally appropriate practice in early childhood programs serving children from birth through age 8. Washington, DC: National Association for the Education of Young Children. Bricker, D. (1995). The challenge of inclusion. Journal of Early Intervention, 19, 179– 194. Bricker, D.D., Bruder, M.B., & Bailcy, E. (1982). Developmental integration of preschool children. Analysis and Intervention in Developmental Disabilities, 2,207– 222. Bricker, D., & Woods-Cripe, J.J. (1992). An activity-based approach to early intervention. Baltimore: Paul H.Brookes. Bristol, M.M., Cohen, D.J., Costello, E.J., Denckla M., Eckberg, T.J., Kallen, R., Kraemer, H.C., Lord. C., Maurer. R., Mellvane, W.J., Minshew, N., Sigman, M., & Spence. M.A. (1996). State of the science in autism: Report to the National Institutes of Health. Journal of Autism and Developmental Disorders, 26, 121–154. Cohen, D.J., & Volkmar, F.R. (Eds.). (1997). Handbook of autism and pervasive developmental disorders (2nd ed.). New York: Wiley. Dawson, G., & Osterling, J. (1997). Early intervention in autism. In M.J.Guralnick (Ed.), The effectiveness of early intervention (pp. 307–326). Baltimore: Paul H. Brookes. Dunlap, G., & Fox, L. (1996). Early intervention and serious problem behaviors. In L.K. Koegel, R.L.Koegel, & G.Dunlap (Eds.), Positive behavioral support: Including people with difficult behavior in the community. Baltimore: Paul H.Brookes. Fenske, E.C., Zalenski, S., Krantz, P.J., & McClannahan, L.E. (1985). Age at intervention and treatment outcome for autistic children in a comprehensive intervention program. Analysis and Intervention in Developmental Disabilities. 5, 49– 58. Ferster, C.B. (1961). Positive reinforcement and behavioral deficits in autistic children. Child Development, 32, 312–345. Ferster, C.B., & Skinner, B.F. (1957). Schedules of reinforcement. New York: AppletonCentury-Crofts. Fox, L., Dunlap, G., & Philbrick, L.A. (1997). Providing individual supports to young children with autism and their families. Journal of Early Intervention, 21, 1–14. Green, G. (1996). Evaluating claims about treatments for autism. In C.Maurice, G. Green, & S.C.Luce (Eds.), Behavioral interventions for young children with autism: A manual for parents and professionals (pp. 15–27). Austin. TX: Pro-Ed. Greenspan, S.I., & Wieder, S. (1997a). An integrated developmental approach to
Early intervention programs for children with autism
411
interventions for young children with severe difficulties in relating and communicating. In S.I.Greenspan, B.Kalmanson, R.Shahmoon-Shanok, S. Wieder, G. G.Willamson, & M.Anzalone (Eds.), Assessing and treating infants and young children with severe difficulties in relating and communicating (pp. 5–18). Washington DC: Zero to Three. Greenspan, S.I., & Wieder, S. (1997b). Developmental patterns and outcomes in infants and children with disorders in relating and communicating: A chart review of 200 cases of children with autistic spectrum diagnoses. Journal of Developmental and Learning Disorders, 1, 87–141. Greenspan, S.I., & Wieder, S. (1998). The child with special needs: Encouraging intellectual and emotional growth. Reading, MA: Addison-Wesley. Guralnick, M.J. (1990). Major accomplishments and future directions in early childhood mainstreaming. Topics in Early Childhood Special Education, 10(2), 1–17. Hanline, M.F., & Golant, K. (1993). Strategies for creating inclusive early childhood setting. In D.M.Bryant & M.A.Graham (Eds.), Implementing early intervention: From research to effective practice. New York: Guilford Press. Harris, S.L., & Handleman, J.S. (Eds.). (1994). Preschool education programs for children with autism. Austin, TX: Pro-Ed. Jenkins, J.R., Speltz, M.L., & Odom, S.L. (1985). Integrating normal and handicapped preschoolers: Effects on child development and social interaction. Exceptional Children, 52, 7–18. Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217–250. Kazdin, A.E. (1993). Replication and extension of behavioral treatment of autistic disorder. American Journal on Mental Retardation, 97, 377–379. Koegel, R.L., & Koegel, L.K. (1995), Teaching children with autism. Baltimore: Paul H.Brookes. Lord, C., Bristol, M.M., & Schopler, E. (1993). Early intervention for children with autism and related developmental disorders. In E.Schopler, M.E.Van Bourgondien, & M.M.Bristol (Eds.), Preschool issues in autism (pp. 199–221). New York: Plenum Press. Lord, C., & Schopler, E. (1989). Stability of assessment results of autistic and nonautistic language-impaired children from preschool years to early school age. Journal of Child Psychology and Psychiatry, 30, 575–590. Lord, C., & Schopler, E. (1994). TEACCH Services for preschool children. In S.L. Harris & J.S.Handelman (Eds.). Preschool education programs for children with autism (pp. 87–106). Austin, TX: Pro-Ed. Lovaas, I.D. (1981). Teaching developmentally disabled children: The ME book. Baltimore: University Park Press. Lovaas, O.I. (1987). Behavioral treatment and normal educational and intellectual functioning in young autistic children. Journal of Consulting and Clinical Psychology, 55, 3–9. Lovaas, O.I., & Smith, T. (1989). A comprehensive behavioral theory of autistic children: Paradigm for research and treatment. Behavior Therapy and Experimental Psychiatry, 20, 17–29. McEachin, J., Smith, T., & Lovaas, O.I. (1993). Long-term outcome for children with autism who received early intensive treatment. American Journal on Mental
Annual progress in child psychiatry and child development 2000–2001
412
Retardation, 97, 359–372. Melsles, S. (1985). Why are we still asking this question? Topics in Early Childhood Special Education, 5, 1–11. Mesibov, G.B. (1996). Division TEACCH: A collaborative model program for service delivery, training, and research for people with autism and related communication handicaps. In M.C.Roberts (Ed.). Model programs in child and family mental health (pp. 215–230). Hillsdale, NJ: Erlbaum. Mesibov, G.B., & Shea, V. (1998). The culture of autism: From theoretical understanding to education practice. [Online]. Available http://www.rmple.co.uk/eduweb/sites/autism/culture.html National Early Childhood Technical Assistance System [NEC*TAS]. (1997, July, November). Developing state and local services for young children with autism and their families, Meetings in Denver, CO, & Clearwater, FL. Odom, S.L., & Brown, W.H. (1993). Social interaction skills interventions for young children with disabilities in integrated settings. In C.Peck, S.Odom, & D.Bricker (Eds.), Integrating young children with disabilities into community programs (pp. 39– 64). Baltimore: Paul H.Brookes. Olley, J.G., & Gutenlag, S.S. (in press). Autism: Historical overview, definition and characteristics. In D.E.Zager (Ed.), Autism: Identification, education, and treatment (2nd ed). Hillsdale, NJ: Erlbaum. Ozonoff, S., & Catheart, K. (1998). Effectiveness of a home program intervention for young children with autism. Journal of Autism and Developmental Disorders, 28, 25– 32. Peeters, J. (1997). Autism: From theoretical understanding to educational intervention. London: Whurr Publishers. Piaget, J. (1952). The origins of intelligences in children. New York: Norton. Rogers, S.J. (1996). Brief report: Early intervention in autism, Journal of Autism and Developmental Disorders, 26, 243–246. Rogers, S.J. (in press). Empirically supported treatment for young children with autism. Journal of Clinical Child Psychology. Rogers, S.J., & Pennington, B.F. (1991). A theoretical approach to the deficits in infantile autism. Development and Psychopathology, 3, 137–162. Schopler, E., Relchler, R.J., & Renner, B.R. (1985). Childhood Autism Rating Scale (CARS). New York: Irvington. Schopler, E., Short, A., & Mesibov, G. (1989). Relation of behavioral treatment to “normal functioning”: Comment on Lovaas. Journal of Consulting and Clinical Psychology, 57, 162–164. Siegel, B. (1996). The world of the autistic child: Understanding and treating autistic spectrum disorders. New York: Oxford University Press. Skinner, B.J. (1968). The technology of teaching. New York: Appleton-Century-Crofts. Smith, T., & Lovaas, O.I. (1998). Intensive and early behavioral intervention with autism: The UCLA young autism project. Infants and Young Children, 10(3), 67–78. Strain, P.S., & Cordisco, L.K. (1994). LEAP preschool. In S.L.Harris & J.S. Handelman (Eds), Preschool education programs for children with autism (pp. 225–244). Austin, TX: Pro-Ed.
Early intervention programs for children with autism
413
Strain, P.S., Hoyson, M., & Jamieson, B. (1985). Normally developing preschoolers as intervention agents for autistic-like children: Effects on class deportment and social interaction. Journal of the Division for Early Childhood, 9, 105–115. Strain, P.S., Kohler, F.W., & Goldstein, H. (1996). Learning experiences: An alternative program: Peer-mediated interventions for young children with autism. In E. Hibbs & P.Jensen (Eds.), Psychosocial treatments for child and adolescent disorders (pp. 573– 586). Washington, DC: American Psychological Association. Warren, S.F., & Kalser, A.P. (1988). Research in early language intervention. In S.L. Odom & M.B.Kames (Eds.), Early intervention for infants and children with handicaps: An empirical base (pp. 89–108). Baltimore: Paul H.Brookes. Wolery, M., Balley, D.B., & Sugai, G.M. (1988). Effective teaching: Principles and procedures of applied behavior analysis with exceptional students. Boston: Allyn and Bacon.
PART V: TREATMENT ISSUES
23 Treatment for Sexually Abused Children and Adolescents Karen J.Saywitz, Anthony P.Mannarino, Lucy Berliner, and Judith A.Cohen
The authors review research demonstrating the variable effects of childhood sexual abuse, the need for intervention, and the effectiveness of available treatment models. The well-controlled treatment-outcome studies reviewed do not focus on sensationalistic fringe treatments that treat sexually abused children as a special class of patients. Instead, studies demonstrate empirical evidence for extending and modifying treatment models from mainstream clinical child psychology to sexually abused children. The authors propose a continuum of interventions to meet the needs of this heterogeneous group. Interventions range from psychoeducation and screening, to shortterm, abuse-focused cognitive—behavioral therapy with family involvement, to more comprehensive long-term plans for multiproblem cases. Last discussed are gaps in the research and suggestions for future research to address the pressing dilemmas faced by clinicians and policymakers.
INTRODUCTION We examine key issues in the treatment of sexually abused children1 and their families. After reviewing the principal findings on the effects of childhood sexual abuse, we describe the available research on treatment efficacy and discuss a number of the difficult questions still facing practitioners, researchers, and policy makers today.
WHAT ARE THE EFFECTS OF CHILD SEXUAL ABUSE? The bulk of past research on childhood sexual abuse has suggested two important findings. First, the impact of child sexual abuse is highly variable. Some children show 1
References to children are intended to include both children and adolescents except when specific age ranges are provided.
Annual progress in child psychiatry and child development 2000–2001
416
no detectable negative effects; others show highly adverse reactions with severe psychiatric symptomatology (e.g., Kendall-Tackett, Williams, & Finkelhor, 1993). Second, child sexual abuse is a risk factor for the development of psychiatric disorders and distress in adults, although not all individuals will experience long-term effects (e.g., Fergusson, Horwood, & Lynskey, 1996; Glaser, 1991; Mullen, Martin, Anderson, Romans, & Herbison, 1996; Saunders, Kilpatrick, Hansen, Resnick, & Walker, 1999; Silverman, Reinherz, & Giaconia, 1996; Widom, 1999). Variability in the effects of child sexual abuse is not surprising given the wide range of experiences that constitute sexual abuse and the disparate contexts in which it can occur, ranging from indecent exposure in a park, to kidnapping and rape at knife point, to many years of multiple forms of maltreatment in a chaotic family situation. Findings from both clinical and community samples emphasize that sexually abused children exhibit more symptoms than nonabused children in comparison groups (e.g., Browne & Finkelhor, 1986; Green, 1993; Kendall-Tackett et al., 1993; Mannarino, Cohen, & Gregor, 1989; Wind & Silvern, 1994). Yet, no one symptom characterizes the majority of sexually abused children, and there is no evidence of a single cohesive syndrome resulting from child sexual abuse. Although no syndrome has been identified, studies do suggest that more than 50% of sexually abused children meet partial or full criteria for posttraumatic stress disorder (PTSD; McLeer, Deblinger, Atkins, Foa, & Ralphe, 1988; McLeer, Deblinger, Henry, Orvashel, 1992). One of the impediments to research has been that diagnostic criteria for PTSD are not sufficiently sensitive to developmental factors, especially the ways in which younger children exhibit effects of trauma. Moreover, such symptoms can be difficult to measure, and available instruments are of limited use with children. Hence, the disorder may be underdiagnosed among children (American Academy of Child and Adolescent Psychiatry [AACAP], 1998). Most researchers underscore the need for longitudinal designs to disentangle the separate effects of sexual abuse (e.g., PTSD symptoms and sexual behavior problems) from the effects of difficulties that predate the abuse and/or continue thereafter (e.g., depression and aggressivity; see, e.g., Briere, 1992; Green, 1993; Stevenson, 1999). In a comprehensive review of 45 studies, KendallTackett et al. (1993) concluded that sexual abuse accounted for 15% to 45% of the variance. Investigators are beginning to tease apart the multiple pathways by which abuse leads to adverse consequences and the underlying psychological mechanisms responsible (e.g., Coffey, Leitenberg, Henning, Turner, & Bennett, 1996). Evidence is beginning to accumulate to demonstrate that the experience of sexual abuse itself makes an independent contribution to later symptoms of PTSD (e.g., Wind & Silvern, 1994). As with any potentially traumatic experience, the effects depend not only on the characteristics of the incident but also on a given child’s vulnerability and resilience. Effects are influenced by the child’s level of preabuse functioning (e.g., temperament, neurodevelopmental reactivity, attachment status) and by the existence of risk and protective factors, including the social resources (e.g., family functioning), emotional resources (e.g., mental health of nonoffending parents), and financial resources (e.g., access to treatment) available to help the child cope with the abusive incident or incidents. Sometimes abuse exacerbates preexisting problems; sometimes abuse
Treatment for sexually abused children and adolescents
417
overwhelms children who have been functioning reasonably well because sequelae unseat protective factors (e.g., children are relocated because of divorce or foster placement, which results in the loss of attachment figures, friends, mentors, and activities that provided recognition, like athletics). Two thirds to one half of sexually abused children appear to improve over time, but many either do not improve or deteriorate (e.g., Kendall-Tackett et al., 1993; Oates, O’Toole, Lynch, Stern, & Cooney, 1994). In empirical studies, most of the children who present as asymptomatic remain symptom free (70%), although some (30%) do develop symptoms later (Kendall-Tackett et al., 1993). The proportion of sexually abused children who present with no detectable symptoms varies across studies from 21% to 49%. In some cases, children may be experiencing symptoms not measured by investigators, or children may be at an early stage before symptoms emerge. Researchers have found evidence of a sleeper effect in severely abused children, with more serious symptoms not manifesting themselves until a year after disclosure (Mannarino, Cohen, Smith, & Moore-Motily, 1991). In other cases, asymptomatic children may represent a particularly resilient group that copes well and never shows symptoms. Other children may not display symptoms of trauma simply because the event was a relatively minor incident that was not experienced as traumatic, although it was exploitative and illegal. Early studies of long-term effects indicated that child sexual abuse is a major risk factor for a variety of problems in adult life, especially those studies that focused on the effects of more serious abuse, on female samples, and/or on patients in treatment, as well as those that used objective and retrospective self-report measures (e.g., Berliner & Elliott, 1996; Browne & Finkelhor, 1986; Coffey et al., 1996; Finkelhor, 1990; Green, 1993; Mullen et al., 1996; Stevenson, 1999). However, conclusions have been restricted by the potential biases inherent in retrospective data. Recently, studies with enhanced designs have provided further support for the notion that child sexual abuse is a risk factor for adult difficulty. These include prospective studies of adults with independently confirmed histories of abuse (e.g., Widom, 1999), studies with repeated childhood testings of psychiatric symptomatology (e.g., Fergusson et al., 1996), and studies using national probability samples (Saunders et al., 1999).2 Some children experience events legally defined as sexual abuse but show no immediate symptoms requiring psychological treatment; nonetheless, these children may benefit from psychoeducational efforts to prevent future victimization and from periodic reassessment to check for sleeper effects, as discussed in the Asymptomatic Children section. In contrast, other children and adolescents experience serious abuse and serious abuse-related sequelae that clearly cause them to need treatment. Currently, there is no reliable means of predicting, in individual cases, which children will have persistent symptomatology or will develop symptoms later and which children require no or minimal intervention, although some predictive variables have been identified. In aggregate, studies have tended to describe four sizable groups of children to be considered for intervention: 1. Some children have no detectable difficulties on standardized measures of child behavior problems. Kendall-Tackett et al. (1993) estimated this group to be about one third of children studied. 2. Some children have a few symptoms that do not reach clinical levels of concern
Annual progress in child psychiatry and child development 2000–2001
418
(e.g., emotional distress, anxiety, self-esteem, or identity difficulties) or have behavior problems that reach clinical levels but are not as severe as in the general clinical population (Cohen & Mannarino, 1988; Einbender & Friedrich, 1989; Gomes-Schwartz, Horowitz, & Cardarelli, 1990; Mannarino et al., 1989; Tong, Oates, & McDowell, 1986; Wolfe, Gentile, & Wolfe, 1989). 3. Some children have serious psychiatric symptoms, such as depression (e.g., Shapiro, Leifer, Martone, & Kassem, 1990), anxiety (e.g., Kolko, Moser, & Weldy, 1988), sexualized behavior (e.g., Friedrich et al., 1992; Gale, Thompson, Moran, & Sack, 1988; Kolko et al., 1988), substance abuse (e.g., Hibbard, Ingersoll, & Orr, 1990; Singer, Petchers, & Hussey, 1989), aggressivity (e.g., Friedrich, Beilke, & Urquiza, 1987), self-esteem or identity difficulties (e.g., Cavaiola & Schiff, 1989; Hotte & Rafman, 1992; Wozencraft, Wagner, & Pellegrin, 1991), shame and cognitive impairments or distortions (Einbender & Friedrich, 1989), and isolated posttraumatic symptoms, such as flashbacks, nightmares, and repetitive play (e.g., Conte & Schuerman, 1987; McLeer et al., 1992; Wolfe et al., 1989). 4. Some children meet full criteria for psychiatric disorders, most notably PTSD, major depression, overanxious disorder, and sleep disorder (e.g., 2
For example, in one longitudinal study of a community sample of abused children, researchers found high rates of depression, substance abuse, and PTSD among the participants as young adults, with suicidal ideation specifically associated with child sexual abuse in female participants (Silverman et al., 1996).
McLeer et al., 1988). In addition, comorbidity is a significant problem with traumatized children and adolescents (AACAP, 1998). Studies of clinical populations estimate that 55% of children referred for treatment have more than one diagnosis (Target & Fonagy, 1996).
IS TREATMENT EFFECTIVE? Obstacles to Evaluating the Efficacy of Child Treatment Researchers have faced a number of obstacles to investigating whether psychological intervention is effective with children and adolescents. First, it has been difficult to identify which symptoms to target with treatment and monitor for change because children are not completely reliable informants about their own mental states, as their self-awareness, vocabulary, reasoning, and metacognitive abilities are still developing. Using parents’ and teachers’ reports may facilitate problem identification; however, these three sources of information often diverge greatly from one another, thus producing conflicting pictures of the problem (e.g., Kolko & Kazdin, 1993). In cases of sexual abuse, children are not typically referred for treatment in the usual manner; that is, they are not referred because of emotional or behavioral difficulties. Instead, they are referred because it is discovered that they have experienced a sexually abusive event. Hence, this is a diverse set of children, varying in age, history, and presentation, and this diversity impairs the researcher’s ability to administer distinctive
Treatment for sexually abused children and adolescents
419
standardized treatments and to use the same outcome measures for all children. Measures need to have different norms and different forms for children who differ in age, gender, vocabulary level, reading and writing ability, and socioeconomic status. Measures for the disorders most common among sexually abused children (e.g., PTSD) have not been developed adequately to detect symptoms in young children (AACAP, 1998). The heterogeneity of this population complicates research design, requiring a great many replication studies on children of different ages, genders, symptom patterns, and family contexts. Additionally, improvement is dependent not only on the efficacy of the treatment and the nature and severity of the patient’s impairment, but also on the functioning of the adults on whom the child depends. The mental health of parents, parents’ marital conflict, family functioning, the presence of stressful life events, the family’s socioeconomic status, and community and cultural factors influence the degree and maintenance of improvement (Kazdin & Weisz, 1998). In cases of child sexual abuse, parental belief and support have been found to play a significant role in treatment outcome. In addition, researchers need to consider unique contextual factors, such as the inclusion of intrafamilial perpetrators in treatment plans and outcome studies where reunification is a goal. Also problematic is the study of children living in multiproblem families. These children may have been exposed to community and domestic violence and other forms of maltreatment, and although sexual abuse is part of their history, it is not the presenting complaint. High spontaneous remission rates also make it difficult to evaluate treatment outcome. Experimental designs need to assess whether treatment quickens recovery and reduces the risk of reoccurrence, even if some children’s distress will remit on its own. In addition, the developmental course of childhood disorders does not always involve a stable set of symptoms. Changing symptom patterns over the course of development complicate research designs. The referral symptom may be improved, but the same underlying difficulty may manifest itself differently at later stages of development. Target and Fonagy (1996) also noted that efficacy cannot be evaluated without attention to the match between the treatment modality and the child’s capabilities and circumstances. Despite indications that family therapy may be best for a particular problem, it is not possible to conduct family therapy when children are in changing foster care placements. Certain therapies can be applied only when the circumstances and the child’s capabilities allow them. Studies of Treatment Outcome with Children and Adolescents Given the obstacles to highly controlled research on treatment outcomes and the very small number of well-controlled studies involving sexually abused children, recommendations for service delivery demand reliance on a broad literature, namely research on interventions with children generally. Moreover, many of the symptoms displayed by abused children, such as depression, anxiety, and aggressivity, are also exhibited by children who have not been sexually abused. Hence, the effectiveness of the available interventions for reducing these symptoms is also at issue (Stevenson, 1999). On the one hand, some evidence suggests that victims of child sexual abuse are less
Annual progress in child psychiatry and child development 2000–2001
420
responsive to some generic treatment strategies as adults than individuals without a history of sexual abuse (Holmes, 1995). On the other hand, once the immediate abusefocused work with the family is complete, researchers do not know whether the moderate and long-term difficulties of the child are all that different from those of other children in need of treatment for similar symptoms (Stevenson, 1999). Recently, the literature on children’s treatment outcomes has enjoyed some keen advances in experimental design and statistical methods that have laid the groundwork for several large meta-analytic studies of hundreds of fairly wellcontrolled investigations. These meta-analyses have indicated that psychosocial therapies are effective with children and adolescents and are more effective than the passage of time alone (e.g., Kazdin, Bass, Ayers, & Rodgers, 1990; Kazdin & Weisz, 1998; Weisz, Weisz, Alicke, & Klotz, 1987; Weisz, Weisz, Han, Granger, & Morton, 1995). Effect sizes have varied by presenting problem, with greater success for phobic, somatic, and anxiety symptoms than for symptoms related to global adjustment or personality characteristics (e.g., Casey & Berman, 1985). Most summaries of the field now echo the words of Peter Fonagy (1998): The era of generic therapies is over. No treatment can be equally applicable without modification to every disorder…. Nonspecific, poorly structured treatments, such as generic counseling, nonfocused dynamic therapy and a variety of experiential therapies are unlikely to be effective with severe presentations, (p. 133) Hence, the focus of current research is on determining what forms of treatment work best for which groups of children (Target & Fonagy, 1996). A detailed discussion of the evolving criteria for validating treatments is beyond the scope of this article. Let it suffice to say that most reviews of the empirical literature find the following treatments to be efficacious or possibly efficacious for use with children and adolescents: cognitive-behavioral therapy (CBT) for childhood anxiety, coping skills training for childhood depression, and parent management training based on behavioral techniques and cognitive problem-solving training for externalizing behavior problems (e.g., aggression). (See Chambless & Hollon, 1998, or Kazdin & Weisz, 1998, for discussions of efficacy criteria.) For the most part, tests of these interventions have demonstrated positive outcomes that have been replicated by different investigative teams. Generally speaking, studies support behavioral therapy and CBT over nonbehavioral therapies (see Kazdin & Weisz, 1998, for a thorough review). This does not mean that behavioral approaches are best for all types of children and all types of problems. These approaches may enjoy the greatest empirical support in part because they are the most frequently studied. They are short term and are among the easiest to manualize, standardize, and therefore utilize in well-controlled treatment trials. The kinds of nonbehavioral approaches that are most prevalent in clinics have not been well evaluated (e.g., family therapy, brief or focused psychodynamic therapies, structured group treatment). Often the active ingredients are difficult to manualize and standardize (e.g., transference). Moreover, some of the changes such interventions seek to achieve (e.g.,
Treatment for sexually abused children and adolescents
421
altered family-interaction patterns or intrapsychic changes) do not lend themselves to the kinds of standardized tests currently available to measure outcome in terms of symptom relief. Most studies have focused on symptom reduction, neglecting the impact of adaptive real-world functioning and development (Kazdin & Weisz, 1998). The bulk of the studies supporting behavioral and cognitive-behavioral techniques have taken place in university settings, and researchers have had difficulty transferring benefits to community clinics where children may present with more severe symptoms (Hoagwood & Hibbs, 1995; Kazdin & Weisz, 1998; Kendall & Southam-Gerow, 1995). Still, the past 25 years of research have produced substantial support for the efficacy of professional behavioral and cognitive-behavioral interventions with children and adolescents, especially when children present with anxiety-related symptoms, depression, or behavior problems. Outcome Studies of Child Sexual Abuse Treatment Reviews of studies using quantitative outcome measures began to appear only in the past decade (Beutler, Williams, & Zetzer, 1994; Finkelhor & Berliner, 1995; Green, 1993). By and large, studies have focused on abuse-specific therapies that have used cognitivebehavioral techniques in conjunction with psychoeducational interventions, coping-skills training, and family involvement. The major emphasis on cognitive-behavioral and behavioral techniques, specific rather than generic therapies, and the rapid push toward randomized clinical trials has clearly been catalyzed by the conclusions of the general child-treatment literature just discussed. Finkelhor and Berliner (1995) conducted a thorough review of studies that used quantitative measures and sizable numbers of participants prior to 1995. They located only 29 studies worthy of inclusion. The largest number of these studies used simple comparisons of children at two times during professional intervention. All of these studies showed that the treated sexually abused children improved significantly over time; however, these types of pretest-posttest designs without control groups have inherent limitations. It is unclear whether improvement is due to treatment, to the passage of time, or to some other factor outside the treatment. Moreover, studies have found that, as a group, sexually abused children improve over time whether or not they receive treatment (e.g., Gomes-Schwartz et al., 1990). There were a few studies that involved quasi-experimental designs, used large samples of nearly 100 children, and/or attempted to deal with design problems in other ways (e.g., self-comparisons; see, e.g., Deblinger, McLeer, & Henry, 1990; Lanktree & Briere, 1995). These studies suggested that not all problems and not all sexually abused children responded to treatment. Some problems, especially externalizing symptoms as measured on the Child Behavior Checklist (e.g., aggressiveness) and sexualized behavior under certain circumstances were resistant to change. The first few controlled studies using randomized assignment to treatment conditions produced somewhat mixed results. However, not all studies had enough participants or powerful enough designs to detect possible differences, and in some cases it was difficult to ensure that the treatments being compared were distinct (Baker, 1987; Berliner & Saunders, 1996; Hyde, Bentovim, & Monck, 1995; Perez, 1988). For example, Berliner and Saunders (1996) compared the effects of a stress-inoculation training module
Annual progress in child psychiatry and child development 2000–2001
422
embedded in standard abuse-focused treatment with the effects of standard treatment alone. In a study of 100 sexually abused children, they did not find significant differences in anxiety symptoms between treatment groups. One reason postulated for the lack of differences was that the treatment conditions were not different enough from each other in ways that mattered. Also, the standard treatment may have been sufficient, given that children did not present with high levels of pretreatment anxiety. The most recent wave of studies has begun to demonstrate statistically significant benefits. These studies used randomized samples and trials, control groups, standardized instruments, manualized treatments, and adherence or fidelity procedures. Usually some form of abuse-specific CBT was compared with a more nondirective approach or a standard community treatment. Deblinger, Lippman, and Steer (1996) compared a 12week program of abusefocused CBT provided to children only, to parents only, or to both children and parents with standard community care for a total of 100 sexually abused children ages 7 to 13. Results indicated that all groups improved on PTSD symptoms; however, CBT provided directly to the children resulted in significantly greater improvement. Furthermore, providing CBT to the nonoffending parents resulted in significantly more improvement in the children’s depressive symptoms, as well as improvement in the parenting skills of the participating parents. These results support abuse-focused professional intervention with both children and nonoffending parents. Celano, Hazzard, Webb, and McCall (1996) randomly assigned 32 sexually abused girls and their parents to an eight-session CBT intervention or to a nonspecific treatment. They found that the CBT group reported significantly more improvement in parental support for the children, significantly fewer parental expectations that the abuse would have negative effects on the children, and less self-blame among parents. However, children in this group did not show more improvement on PTSD symptoms than the control group. Researchers noted that the most talked about topic even in the control condition was the sexual abuse and the children’s associated reactions. This may have reduced differences between treatment conditions. Also, the children’s PTSD symptoms prior to treatment were not severe, and again, the more structured, manualized approach may not have been necessary. However, the results highlight the added value of parents’ involvement in the treatment of sexually abused girls. Cohen and Mannarino (1996b, 1998b) conducted two studies of sexually abused children that compared CBT to nondirective supportive therapy. Both of these studies included intervention with the nonoffending parent in each treatment condition. In the first study (Cohen & Mannarino, 1996b), 67 3- to 7-year-olds and their parents were randomly assigned to either abuse-specific CBT or nondirective supportive therapy composed of nonspecific play therapy for the children and supportive counseling for the parents. Children provided with abuse-focused CBT had significantly greater improvement in PTSD symptoms, sexually inappropriate behaviors, and internalizing and externalizing symptoms compared with those receiving nondirective supportive therapy. These differences were sustained at a 1-year follow-up (Cohen & Mannarino, 1997). These findings strongly support abuse-focused CBT with preschoolers and their parents. In a study of 49 older children ages 7 to 14, those who were provided with abusefocused CBT experienced significantly greater improvement in depression and social competence than those receiving nondirective supportive therapy (Cohen & Mannarino,
Treatment for sexually abused children and adolescents
423
1998b). However, as in Celano et al.’s (1996) study, the groups did not differ in PTSD symptoms. Again, it could be that when PTSD is not severe and children are older, they spontaneously discuss the abuse and their associated feelings and attributions even in the control condition. However, this study did find that improvement in older sexually abused children’s depressive symptoms was associated with abuse-specific CBT. Additionally, recent studies have identified factors that can exacerbate or ameliorate the psychological impact of sexual abuse and treatment outcome, most notably family factors and children’s attributions. Symptom development and treatment response are influenced by parents’ emotional distress related to the abuse, parents’ support of their children, and children’s attributions regarding variables such as locus of control (Cohen & Mannarino, 2000; Mannarino & Cohen, 1996a, 1996b). In a study that included parent report measures, Mannarino and Cohen (1996c) found that ratings of family cohesion, adaptability, and the intensity of parents’ reactions to the abuse were significantly related to ratings of children’s behavior problems. In a preschool population, parents’ emotional distress about the abuse strongly predicted treatment outcome initially (Cohen & Mannarino, 1996a), but at the 1-year follow-up, parents’ support was a stronger predictor of positive outcome (Cohen & Mannarino, 1998a). In a study of older children, parents’ support and the children’s abuserelated cognitions most strongly predicted treatment outcome (Cohen & Mannarino, 2000). These findings reiterate the importance of parental involvement to treatment outcome. In summary, the existing studies of the outcomes of treatment for children who have been sexually abused suggest that many of the symptoms associated with child sexual abuse can be responsive to professional intervention, although symptom changes are influenced by other factors as well (e.g., intensity of parental reaction to abuse, children’s attributions, and family adaptability). Often, these factors are also targets of treatment. As this review demonstrates, however, the number of treatment outcome studies that have randomly assigned children to treatment conditions has been quite small, and these studies have not been sufficiently replicated by different investigative teams. Still, results consistently favor abuse-specific CBT over the other forms of treatment to which it has been compared, although abuse-specific CBT did not always per-form as expected (i.e., it did not always reduce levels of posttraumatic stress).3 Also, there is some evidence that behavioral interventions with parents were required to decrease externalizing behaviors (aggression) or sexualized behaviors in young children. When considered in conjunction with the literature on child-treatment outcomes as a whole, these results are bolstered by findings that support the efficacy of behavioral and cognitive-behavioral treatments when children present with the kinds of symptoms often seen in sexually abused children, namely, anxiety-related symptoms, depression, or behavioral problems. 3
Inconsistent results on PTSD measures need to be further investigated. As mentioned, researchers still have a good deal of difficulty reliably measuring PTSD symptoms in children, given the specificity of some of the PTSD symptoms in children who have been sexually abused.
The Need for Additional Research
Annual progress in child psychiatry and child development 2000–2001
424
The empirical support for abuse-specific CBT must be viewed in the context of a dearth of information about the efficacy of other treatments. These treatments, prevalent in the field and based on well-articulated theoretical frameworks, have not been adequately tested. Victim groups, family therapy, selfhelp groups involving offenders (e.g., Parents United), and eye-movement desensitization and reprocessing are among these modalities.4 To test these other forms of treatment, however, therapists will need to carefully describe these therapies to inform investigations. In addition, researchers will need to develop global outcome measures of real-world functioning and will need to test treatment effectiveness in community clinics where symptom and treatment-use patterns may differ from those of research participants (e.g., Horowitz, Putnam, Noll, & Trickett, 1997). Definitive studies to determine which forms of intervention are best suited for which groups of sexually abused children have yet to be conducted.5 The influences of moderator variables, risk and protective factors, and preceding and continuing background factors have not been disentangled. The components of abuse-specific CBT have not been tested independently; hence, it is difficult to know whether individual components, such as psychoeducation, would prevent future victimization, deterioration, or late-emerging symptoms if they were offered to children with few, minor, or no symptoms. Long-term outcome has rarely been investigated; hence, researchers know little about the role of treatment in relapse prevention or in forestalling symptoms that would otherwise appear. Similarly, developmental issues in treatment outcome have rarely been addressed, despite the obvious fact that sexually abused children are a heteroge-neous group. These children differ in key attributes, including language, cognitive, and emotional skills and sexual maturity, that are necessary for full participation in many forms of treatment. Generalization from the available studies requires that researchers design studies with multiple age groups to investigate the relative benefits of different treatments for children at different ages. Finally, one of the most pressing issues in the field remains the need to study children who initially present with no or few symptoms. If they are included in outcome studies, they will make it more difficult to detect positive results when positive results exist, because asymptomatic children cannot show improvement. These children probably need to be studied as a distinct group to determine whether early intervention does in fact prevent deterioration, lateemerging symptoms, and further victimization. Informed policy 4
We located only two relevant studies with sexually abused children. In both studies, researchers found almost no group differences when comparing family therapy with family therapy plus group therapy (Hyde et al., 1995) or when comparing group to individual play therapy (Perez, 1988). 5 A multisite study of 240 sexually abused children is currently underway to address some of these questions (Cohen, Deblinger, & Mannarino, 1997).
Treatment for sexually abused children and adolescents
425
decisions about who should be offered what kinds of services demand a better understanding of how asymptomatic children who have been sexually abused respond to the abuse and to treatment over time.
HOW SHOULD SEXUALLY ABUSED CHILDREN AND ADOLESCENTS BE TREATED? Given the diversity of the effects of child sexual abuse, no single type of intervention is likely to be applicable or effective for all sexually abused children. Often, the treatment of sexually abused children and their families is complex. Treatment plans need to be individualized on the basis of the clinical presentation of the child and the context in which treatment will proceed. Multimodal treatment (individual, family, group, pharmacological) and different levels of care (e.g., outpatient, partial, or inpatient) may be required for different children or for the same child at different times (AACAP, 1998). In addition, working with caretakers in one form or another appears to be essential. Including parents in treatment enables them to manage externalizing symptoms with behavioral strategies, to monitor children’s symptoms, to develop strategies for preventing revictimization, and to normalize family functioning. Involvement in treatment helps parents control their own distress and reframe their own attributional errors so that they can support the child’s coping. Standards for practice in cases of child sexual abuse are no different from those used for other types of cases. Fundamental principles of screening, assessment, and treatment planning apply. It bears repeating, however, that there is little empirical support for generic therapies applied indiscriminately to all cases regardless of the referral question. Hence, as with referrals for traditional psychiatric disorders or other potentially traumatic events, clinicians who work with sexually abused children need to think strategically. Specific symptoms need to be targeted with specific strategies. Abuse-Specific Treatment One reason that abuse-specific CBTs are potentially helpful is that they incorporate wellestablished treatment strategies to target specific symptoms. Interventions target the chief symptoms of post-traumatic distress (e.g., reexperiencing the event with intrusive thoughts or flashbacks, avoidance of reminders, and hyperarousal). Anxiety and avoidance are targeted with gradual exposure and desensitization, stress inoculation and relaxation training, and interruption and replacement of upsetting thoughts to regain control over thoughts and feelings. Depressive symptoms are targeted with coping-skills training and correction of cognitive distortions. Behavior problems that interfere with functioning are targeted with conventional behavior-management strategies. Given the limited but consistent support for abuse-focused behavioral and cognitive-behavioral interventions, those who work with sexually abused children will want to familiarize themselves with this approach as one of the methods available in their repertoire of therapeutic options (see Cohen, Mannarino, Berliner, & Deblinger, 2000). One assumption underlying the need for abuse-specific treatment is that these children
Annual progress in child psychiatry and child development 2000–2001
426
have been involved in a potentially traumatic and inappropriate sexual experience that warrants a treatment focus not only on amelioration but also on prevention—prevention of both further victimization and later behavior problems (e.g., substance abuse, promiscuity, or depression in adolescence). This is not an unreasonable assumption given the rates at which children in high-risk situations are abused multiple times and the fact that a portion of children who initially present with no measurable symptoms do develop difficulties later. However, the specific methods used to implement abuse-specific treatment must be carefully considered. A dilemma is created by the fact that interventions with the most empirical support involve exposure-based treatment. When abuse is the source of extreme anxiety and avoidance triggered by remembering, some direct discussion of the abusive event or events is indicated for exposure and desensitization to be effective. When the abuse is the source of distorted cognitions that underlie depression, some direct discussion of the abusive event is often indicated for cognitive restructuring to be effective. However, if the discussion is not carefully conducted, it may contaminate the child’s report and the disclosure process. Sometimes it may be appropriate to delay direct discussion of the abusive event with a child until the abuse is substantiated by the local child protection service system. Much potentially therapeutic work can occur early in treatment without extensive discussion of the specifics of the abuse. This work can include crisis intervention to deal with police or out-of-home placement and symptom reduction through pharmacological agents for depression or anxiety. However, final resolution within the legal system is often a matter of years, not weeks or months. If intervention is delayed for too long, symptoms can exacerbate or become chronic and resistant to treatment. Withholding empirically supported treatment from children who exhibit serious symptoms of posttraumatic stress (e.g., flashbacks, nightmares, phobic avoidance) raises ethical questions to be carefully weighed. It is clear that therapists need to be aware of the legal implications of their interventions. Caution and forethought must be exercised before discussing the event with a child to minimize the potential for distortion. Familiarity with the research on young children’s suggestibility will highlight the dangers of telling rather than asking young children what happened (Saywitz & Lyon, in press). Children need to be able to describe events in their own words. Currently, researchers are developing and testing innovative techniques that minimize the potential for distortion when talking to children about past experiences (e.g., narrative elaboration training, which was developed by Saywitz & Snyder, 1996; see also Camparo, Wagner, & Saywitz, in press). In addition, when therapists transform into interrogators, they can find themselves in an ethical quagmire created by the dual roles. Most professional organizations recommend that forensic evaluations be undertaken separately by specially trained professionals. There is probably no way to escape the inherent tensions among the goals of mental health recovery, child protection, due process for the accused, and justice, all of which need to be carefully balanced on a case-by-case basis. Caution, interdisciplinary coordination, and consultation with colleagues is often required.
Treatment for sexually abused children and adolescents
427
Asymptomatic Children Whether to allocate limited treatment resources to children who present with no detectable symptoms is a difficult issue. Unfortunately, little research is available to guide decision making. However, it is routine to provide debriefing and psychoeducation to all exposed children in the aftermath of natural disasters or school violence, even though most of the children display no overt symptoms. This is widely considered an important preventive intervention, as well as an opportunity to screen for risk factors. Just like children exposed to other potentially traumatic events, sexually abused children may very well benefit from efforts to prevent them from developing misperceptions, unrealistic fears, and maladaptive coping patterns that can grow into problems of clinical proportions. Also, providers will be in an excellent position to identify potential risk factors, as well as to assess for symptomatology and impaired functioning. Abuse characteristics known to be associated with elevated risk (e.g., violence, penetration, longer duration), negative attributions, and parental distress can be evaluated. In addition, children can be assessed for risk factors that are not specific to the abuse, such as family dysfunction, prior trauma, an immature or mentally ill caretaker, disability, and so forth. Hence, when children present with no overt symptoms, it may be sufficient to provide psychoeducation, screening, and prevention awareness that could be accomplished with in a few sessions. Psychoeducation can be aimed at prevention of further victimization, normalization, positive self-image, and parent edification. Through psychoeducation, parents can be taught to identify signs of difficulty that can emerge at later developmental stages, which can increase the probability of reevaluation when appropriate. Such limited professional intervention lays the groundwork for later treatment seeking if it becomes necessary. For children with no symptoms but high levels of risk, psychoeducational efforts can be followed by monitoring with periodic reevaluation to detect signs of deterioration or late-emerging symptoms that were not apparent initially, perhaps because of an avoidant coping style or external supports that later gave way. Multi-Problem Cases At the other end of the spectrum are children for whom sexual abuse is only one of many adverse life experiences that need to be addressed. Sexual abuse can last for years and can coccur with other forms of maltreatment, community violence, and other comorbid psychiatric conditions. Sometimes sexual abuse exacerbates already existing developmental delays or potentiates ongoing psychological, behavioral, or interpersonal difficulties. With placement in foster care, some sexually abused children suffer consequences, including the loss of family and friends, that cause additional symptoms distinct from the abusive incident itself. In these complex cases, children’s reactions to the sexually abusive incident or incidents are often not the only or the most pressing issues in need of treatment. Some cases require long-term, multi-faceted intervention strategies, strategies that have not been empirically validated. Interventions aimed at abuse-specific consequences fall short
Annual progress in child psychiatry and child development 2000–2001
428
if they are not part of a more comprehensive treatment plan to address a multitude of difficulties (e.g., attachment disorders, learning disabilities). Although studies show support for short-term approaches (8 to 16 sessions), long-term approaches have not been tested, and there is no clear evidence regarding the proper length of treatment. In the field, some children and parents participate in wide-ranging therapeutic activities that vary over time. For example, children with multiple problems may benefit from short-term abusefocused therapy in conjunction with long-term relationship-based therapies to help them cope with everyday life. However, treatmentoutcome studies have focused primarily on less complicated diagnostic pictures and higher functioning families. The results are insufficient to dictate the most effective ways to organize treatments for multiproblem cases. Reunification and Resolution Cases For children who have been abused by family members, there are special considerations. In some cases, the child and nonoffending parent may wish to reunify with an offending parent or sibling. In other cases, although reunification may not be a consideration, resolution of the abuse experience with the offender nonetheless may be important for the child to fully resolve the impact of the experience, because the offender who is a relative may continue to be a part of the child’s life. Regardless of residency, biological parents and siblings remain significant figures for children. At this time, there are no empirical data regarding the frequency of reunification when sexual abuse has occurred or whether it is helpful or harmful to the victim. There are no data addressing the safest and most helpful methods to achieve successful therapeutic resolution of this very complex psychological circumstance, although helpful clinical descriptions of family-resolution therapies are available (Saunders & Mienig, 2000).6
CONCLUSIONS Research on the efficacy of treatment for sexually abused children has been moving with laudable speed from case studies and seriously flawed designs to studies using random assignment to compare alterative treatments. This literature is poised to make a meaningful and potentially decisive contribution to clinical practice. Current investigations are not focused on unconventional treatments at the fringe but on extending the treatment models in the mainstream of clinical child psychology to the population of sexually abused children and adolescents. The abuse-specific CRTs being studied are based on interventions shown to be efficacious (or probably efficacious) for treating anxiety, depression, and behavior problems in children and adolescents generally. In the beginning of this article, we identified four groups of children to be considered for intervention. These children range from those who show no detectable signs of adverse impact but develop symptoms later to those who meet full criteria for psychiatric disorder when they are first assessed. To accommodate the different levels of care dictated by these different groups, a continuum of interventions is necessary, ranging
Treatment for sexually abused children and adolescents
429
from psychoeducation, to shortterm abuse-focused CRTs with parental involvement, to more comprehensive long-term treatment plans for multiproblem cases. The available research suggests that abuse-specific CRTs are probably effi6
Outcome studies have focused treatment exclusively on children and nonoffending parents. When researchers have examined the impact of perpetrator identity, they have not found it to be a significant moderator of children’s symptomatology or response to treatment (Cohen & Mannarino, 1996a, 2000; Finkelhor, 1990; Mannarino & Cohen, 1996c; Mannarino et al., 1989, 1991).
cacious for alleviating many of the chief symptoms displayed by sexually abused children. Other factors are also influential (e.g., parents’ reactions, family functioning), and it remains difficult to measure some key symptoms and to demonstrate their consistent responsiveness to treatment. Other approaches to treatment that are prevalent in the field have yet to be adequately tested. However, with an aggressive research agenda that addresses many of the issues raised in the foregoing discussion, researchers should be able to identify which interventions are most beneficial for which groups of children under which sets of circumstances. Clinicians who work with sexually abused children will want to familiarize themselves with the most efficacious treatments now available for addressing abuse-specific consequences and will want to incorporate these treatments into standard practices of screening, assessment, and treatment planning. Yet, clinicians can anticipate an expanding knowledge base and the need to continually reconsider and revise their approaches. The path to effective treatment is clear: Rather than polarizing their efforts, researchers and practitioners will need to collaborate to test the long-term effectiveness of different treatment approaches in the community and to deliver optimal services to sexually abused children and their families.
ACKNOWLEDGMENTS Melissa G.Warren served as action editor for this article. Karen J.Saywitz, Department of Psychiatry and Behavioral Sciences, University of California, Los Angeles (UCLA) School of Medicine; Department of Psychiatry, HarborUCLA Medical Center, Torrance, CA. Anthony P.Mannarino and Judith A. Cohen, Department of Psychiatry, MCPHahnemann University School of Medicine; Department of Psychiatry, Allegheny General Hospital, Pittsburgh, PA.Lucy Berliner, Harborview Center for Sexual Assault and Traumatic Stress, University of Washington.
REFERENCES American Academy of Child and Adolescent Psychiatry. (1998). Practice parameters for the assessment and treatment of children and adolescents with posttraumatic stress disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 4S–26S.
Annual progress in child psychiatry and child development 2000–2001
430
Baker, C.R. (1987). A comparison of individual and group therapy as treatment of sexually abused adolescent females. Dissertation Abstracts International, 47, 4319– 4320. Berliner, L., & Elliott, D. (1996). Sexual abuse of children. In J.Briere, L.Berliner, J. Bulkley, C.Jenny, & T.Reid (Eds.), The APSAC handbook on child maltreatment (pp. 51–71). Thousand Oaks, CA: Sage. Berliner, L., & Saunders, B. (1996). Results of a controlled two-year follow-up study. Child Maltreatment, 1, 294–309. Beutler, L.E., Williams, R.E., & Zetzer, H.A. (1994). Efficacy of treatment for victims of child sexual abuse. The Future of Children, 4, 156–175. Briere, J. (1992). Methodological issues in the study of sexual abuse effects. Journal of Consulting and Clinical Psychology, 60, 196–203. Browne, A., & Finkelhor, D. (1986). Impact of childhood sexual abuse: A review of the research. Psychological Bulletin, 99, 66–77. Camparo, L.B., Wagner, J.T., & Saywitz, K.J. (in press). Interviewing children about real and fictitious events: Revisiting the narrative elaboration procedure. Law and Human Behavior. Casey, R.J., & Berman, J.S. (1985). The outcome of psychotherapy with children. Psychological Bulletin, 98, 388–400. Cavaiola, A.A., & Schiff, M. (1989). Self-esteem in abused, chemically dependent adolescents. Child Abuse and Neglect, 13, 327–334. Celano, M., Hazzard, A., Webb, C, & McCall, C. (1996). Treatment of traumagenic beliefs among sexually abused girls and their mothers: An evaluation study. Journal of Abnormal Child Psychology, 24, 1–16. Chambless, D.L., & Hollon, S.D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66, 7–18. Coffey, P., Leitenberg, H., Henning, K., Turner, T., & Bennett, R.T. (1996). Mediators of the long-term impact of child sexual abuse: Perceived stigma, betrayal, powerlessness and self-blame. Child Abuse and Neglect, 20, 447–455. Cohen, J.A., Deblinger, E., & Mannarino, A.P. (1997). Treatment ofPTSD in sexually abused children [Proposal for Grant R10MH55963 funded by the National Institute of Mental Health]. Bethesda, MD: National Institute of Mental Health. Cohen, J.A., & Mannarino, A.P. (1988). Psychological symptom in sexually abused girls. Child Abuse and Neglect, 12, 571–577. Cohen, J.A., & Mannarino, A.P. (1996a). Factors that mediate treatment outcome of sexually abused preschool children. Journal of the American Academy of Child and Adolescent Psychiatry, 34, 1402–1410. Cohen, J.A., & Mannarino, A.P. (1996b). A treatment outcome study for sexually abused preschool children: Initial findings. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 42–50. Cohen, J.A., & Mannarino, A.P. (1997). A treatment study for sexually abused preschool children: Outcome during a one-year follow-up. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 1228–1235. Cohen, J.A., & Mannarino, A.P. (1998a). Factors that mediate treatment outcome of sexually abused preschool children: Six- and 12-month follow-up . Journal of the
Treatment for sexually abused children and adolescents
431
American Academy of Child and Adolescent Psychiatry, 37, 44–51. Cohen, J.A., & Mannarino, A.P. (1998b). Interventions for sexually abused children: Initial treatment findings. Child Maltreament, 3, 17–26. Cohen, J.A., & Mannarino, A.P. (2000). Predictors of treatment outcome in sexually abused children. Child Abuse and Neglect, 24, 983–994. Cohen, J.A., Mannarino, A.P., Berliner, L., & Deblinger, E. (2000). Trauma focused cognitive behavior therapy: An empirical update. Journal of Interpersonal Violence, 15, 1203–1223. Conte, J., & Schuerman, J. (1987). Factors associated with an increased impact of sexual abuse. Child Abuse and Neglect, 11, 201–211. Deblinger, E., Lippman, J., & Steer, R. (1996). Sexually abused children suffering posttraumatic stress symptoms: Initial treatment outcome findings. Child Maltreatment, 1, 310–321. Deblinger, E., McLeer, S.V., & Henry, D. (1990). Cognitive behavioral treatment for sexually abused children: Preliminary findings. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 747–752. Einbender, A.J., & Friedrich, W.N. (1989). Psychological functioning and behavior of sexually abused girls. Journal of Consulting and Clinical Psychology, 57, 155–157. Fergusson, D.M., Horwood, L.J., & Lynskey, M.T. (1996). Childhood sexual abuse and psychiatric disorder in young adulthood: II. Psychiatric outcomes of child sexual abuse. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 1365–1374. Finkelhor, D. (1990). Early and long-term effects of child sexual abuse: An update. Professional Psychology, 21, 325–330. Finkelhor, D., & Berliner, L. (1995). Research on the treatment of sexually abused children: A review and recommendations. Journal of the American Academy of Child and Adolescent Psychiatry, 34, 1408–1423. Fonagy, P. (1998). Prevention, the appropriate target of infant psychotherapy. Infant Mental Health Journal, 19, 124–150. Friedrich, W.N., Beilke, R., & Urquiza, A. (1987). Children from sexually abusive families: A behavioral comparison. Journal of Interpersonal Violence, 2, 391–402. Friedrich, W.N., Grambasch, P., Damon, L., Hewitt, S.K., Koverola, C., Lang, R., & Wolfe, V. (1992). The Child Sexual Behavior Inventory: Normative and clinical comparisons. Psychological Assessment, 4, 303–311. Gale, J., Thompson, R.J., Moran, T., & Sack, W.H. (1988). Sexual abuse in young children: Its clinical presentation and characteristic patterns. Child Abuse and Neglect, 12, 163–170. Glaser, D. (1991). Treatment issues in child sexual abuse. British Journal of Psychiatry, 159, 255–265. Gomes-Schwartz, B., Horowitz, J.M., & Cardarelli, A.P. (1990). Child sexual abuse: The initial effects. Newbury Park, CA: Sage. Green, H. (1993). Child sexual abuse—immediate and long-term effects and intervention. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 890–902. Hibbard, R.A., Ingersoll, G.M., & Orr, D.P. (1990). Behavior risk, emotional risk, and child abuse among adolescents in a nonclinical setting . Pediatrics, 86, 896–901.
Annual progress in child psychiatry and child development 2000–2001
432
Hoagwood, K., & Hibbs, E. (1995). Efficacy and effectiveness in studies of child and adolescent psychotherapy. Journal of Consulting and Clinical Psychology, 63, 683– 687. Holmes, T.R. (1995). A history of childhood abuse as a predictor variable—implications for outcome research. Research on Social Work Practice, 5, 297–308. Horowitz, L.A., Putnam, F.W., Noll, J.G., & Trickett, P.K. (1997). Factors affecting utilization of treatment services by sexually abused girls. Child Abuse and Neglect, 21, 35–48. Hotte, J.P., & Rafman, S. (1992). The specific effects of incest on prepubertal girls from dysfunctional families. Child Abuse and Neglect, 16, 273–283. Hyde, C., Bentovim, A., Monck, E. (1995). Some clinical and methodological implications of a treatment outcome study of sexually abused children. Child Abuse and Neglect, 19, 1387–1399. Kazdin, A.E., Bass, D., Ayers, W.A., & Rodgers, A. (1990). The empirical and clinical focus of child and adolescent psychotherapy research. Journal of Consulting and Clinical Psychology, 58, 729–740. Kazdin, A.E., & Weisz, J.R. (1998). Identifying and developing empirically supported child and adolescent treatments. Journal of Consulting and Clinical Psychology, 66, 19–36. Kendall, P.C., & Southam-Gerow, M.A. (1995). Long-term follow-up of a cognitivebehavioral therapy for anxiety-disordered youth. Journal of Consulting and Clinical Psychology, 64, 724–730. Kendall-Tackett, K.A., Williams, L.M., & Finkelhor, D. (1993). Impact of sexual abuse on children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113, 164–180. Kolko, D.J., & Kazdin, A.E. (1993). Emotional/behavioral problems in clinic and nonclinic children: Correspondence among child, parent, and teacher reports. Journal of Child Psychology and Psychiatry, 34, 991–1006. Kolko, D.J., Moser, J.T., & Weldy, S.R. (1988). Behavioral/emotional indicators of sexual abuse in psychiatric inpatients: A controlled comparison with physical abuse. Child Abuse and Neglect, 12, 529–541. Lanktree, C.B., & Briere, J. (1995). Outcome therapy for sexually abused children: A repeated measures study. Child Abuse and Neglect, 19, 1145–1155. Mannarino, A.P., & Cohen, J.A. (1996a). Abuse related attributions and perceptions, general attributions, and locus of control in sexually abused girls. Journal of Interpersonal Violence, 11, 162–180. Mannarino, A.P., & Cohen, J.A. (1996b). Family related variables and psychological symptom formation in sexually abused girls. Journal of Child Sexual Abuse, 5, 105– 120. Mannarino, A.P., & Cohen, J.A. (1996c). A follow-up study of factors that mediate the development of psychological symptomatology in sexually abused girls. Child Maltreatment, 1, 246–260. Mannarino, A.P., Cohen, J.A., & Gregor, M. (1989). Emotional and behavioral difficulties in sexually abused girls. Journal of Interpersonal Violence, 4, 437–451. Mannarino, A.P., Cohen, J.A., Smith, J.A., & Moore-Motily, S. (1991). Six- and twelve-
Treatment for sexually abused children and adolescents
433
month follow-up of sexually abused girls. Journal of Interpersonal Violence, 6, 494–511. McLeer, S.V., Deblinger, E., Atkins, M.S., Foa, E.B., & Ralphe, D.L. (1988). Posttraumatic stress disorder in sexually abused children. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 650–654. McLeer, S.V., Deblinger, E., Henry, D., & Orvashel, H. (1992). Sexually abused children at high risk for PTSD. Journal of the American Academy of Child and Adolescent Psychiatry, 31, 875–879. Mullen, P.E., Martin, J.L., Anderson, J.C., Romans, S.E., & Herbison, G.P. (1996). The long-term impact of the physical, emotional, and sexual abuse of children: A community study. Child Abuse and Neglect, 20, 7–21. Oates, R.K., O’Toole, B.L., Lynch, D.L., Stern, A., & Cooney, G. (1994). Stability and change in outcomes for sexually abused children. Journal of the American Academy of Child and Adolescent Psychiatry, 33, 945–953. Perez, C.L. (1988). A comparison of group play therapy and individual therapy for sexually abused children. Dissertation Abstracts International, 48, 3079. Saunders, B.E., Kilpatrick, D.G., Hansen, R.F., Resnick, H.S., & Walker, M.E. (1999). Prevalence, case characteristics, and long-term psychological correlates of child rape among women: A national survey. Child Maltreatment, 4, 187–200. Saunders, B.E., & Mienig, M.B. (2000). Immediate issues affecting long-term family resolution in cases of parent-child sexual abuse. In R.Reece (Ed.), The treatment of child abuse: Common ground for mental health, medical, and legal practitioners (pp. 36–53). Baltimore: Johns Hopkins University Press. Saywitz, K.J., & Lyon, T. (in press). Coming to grips with children’s suggestibility. In G.Goodman, J.Quas, & M.Eisen (Eds.), Memory and suggestibility in the forensic interview. New York: Guilford. Saywitz, K.J., & Snyder, L. (1996). Narrative elaboration: Test of a new procedure for questioning children. Journal of Consulting and Clinical Psychology, 64, 1347–1357. Shapiro, J.P., Leifer, M., Martone, M.W., & Kassem, L. (1990). Multimethod assessment of depression in sexually abused girls. Journal of Personality Assessment, 55, 234– 248. Silverman, A.B., Reinherz, H.Z., & Giaconia, R.M. (1996). The long-term sequelae of child and adolescent abuse: A longitudinal community study . Child Abuse and Neglect, 20, 709–723. Singer, M.I., Petchers, M.K., & Hussey, D. (1989). The relationship between sexual abuse and substance abuse among psychiatrically hospitalized adolescents. Child Abuse and Neglect, 13, 319–325. Stevenson, J. (1999). The treatment of the long-term sequelae of child abuse. Journal of Child Psychology and Psychiatry, 40, 89–111. Target, M., & Fonagy, P. (1996). The psychological treatment of child and adolescent psychiatric disorders. In A.Roth & P.Fonagy (Eds.), What works for whom? A critical review of psychotherapy research (pp. 263–320). New York: Guilford. Tong, L., Oates, K., & McDowell, M. (1986). Personality development following sexual abuse. Child Abuse and Neglect, 11, 371–383. Weisz, J.R., Weisz, B., Alicke, M.D., & Klotz, M.L. (1987). Effectiveness of psychotherapy with children and adolescents: A meta-analysis for clinicians. Journal
Annual progress in child psychiatry and child development 2000–2001
434
of Consulting and Clinical Psychology, 55, 542–549. Weisz, J.R., Weisz, B., Han, S.S., Granger, D.A., & Morton, T. (1995). Effects of psychotherapy with children and adolescents revisited: A meta-analysis of treatment outcome studies. Psychological Bulletin, 117, 450–468. Widom, C.S. (1999). Posttraumatic stress disorder in abused and neglected children grown up. American Journal of Psychiatry, 56, 1223–1229. Wind, T.W., & Silvern, L. (1994). Parenting and family stress as mediators of the longterm effects of child abuse. Child Abuse and Neglect, 18, 439–453. Wolfe, V., Gentile, C., & Wolfe, D.A. (1989). The impact of sexual abuse on children: A PTSD formulation. Behavior Therapy, 20, 215–228. Wozencraft, T., Wagner, W., & Pellegrin, A. (1991). Depression and Suicidal ideation in sexually abused children. Child Abuse and Neglect, 15, 505–510.
PART V: TREATMENT ISSUES
24 Neuroleptic Malignant Syndrome in Children and Adolescents Raul R.Silva, Dinohra M.Munoz, Murray Alpert, Ilisse R.Perlmutter, and Jose Diaz
Objective: Neuroleptic malignant syndrome (NMS) is a serious iatrogenic condition. This report reviews the world literature to characterize the syndrome and evaluate factors that promote early detection and effective intervention. Method: The review identified 77 NMS cases (49 males, 27 females, one gender unknown); ages ranged from .9 to 18 years (mean 14.8±3.96). Univariate and multiple regression analyses were applied to 38 variables to identify early signs of the disorder, to identify correlates of outcome, and to evaluate treatments. Results: The duration of NMS spanned from 1 to 119 days. Nine percent of patients died and 20% resolved with serious sequelae. Patients receiving low-potency neuroleptics had a poorer outcome (p=.01). Fever was related to longer duration of illness (p=.03). Anticholinergics and bromocriptine were effective and without fatalities, but dantrolene was not useful in this sample of children and adolescents. Conclusions: Early detection and appropriate interventions appear important in moderating the course and outcome of NMS. (J Am Acad Child Adolesc Psychiatry, 1999, 38(2): 187– 194.) Key Words: neuroleptic malignant syndrome, treatment, prognosis.
INTRODUCTION Delay and Deniker (1968) identified a cluster of adverse effects of antipsychotic medications, including hypertonicity, autonomic instability, fever, and cognitive disturbance, which they named neuroleptic malignant syndrome (NMS). These authors had previously introduced the term neuroleptics to describe the diverse class of therapeutic agents with dual actions: (a) the potential for reducing psychosis; and (b) the potential for inducing a number of extrapyramidal side effects (EPS). If NMS is an aspect of the EPS reaction to neuroleptics, it may not be totally avoidable when the antipsychotic action is required. However, current views hold that the dual actions of neuroleptics can be dissociated and EPS should be avoided, especially when managing a
Annual progress in child psychiatry and child development 2000–2001
436
child’s medication. The action of atypical neuroleptics also argues for a dissociation of antipsychotic and EPS mechanisms, although any possible role of these agents in NMS remains to be elucidated with further experience (Dave, 1995; Singer et al., 1995). Pearlman (1986) reviewed the NMS literature in the general population and cited incidence rates ranging from .5% to 1.4% of individuals exposed to neuroleptics but did not clarify the incidence of NMS in children and adolescents. At least four sets of diagnostic criteria of NMS have been published (American Psychiatric Association, 1994; Caroff, 1980; Levenson, 1985; Pope et al., 1986); they differ somewhat in the minimal number of required signs (Steingard et al., 1992, p. 186), and there is still no consensus on diagnostic requirements. Fever and rigidity are included as cardinal signs in all lists. Seven additional signs and laboratory measures overlap in the 4 criteria sets: elevated creatine phosphokinase (CPK), tachycardia, tachypnea, hypertension, altered consciousness, diaphoresis, and leukocytosis. The frequency of occurrence of these signs differs. Elevated CPK has been identified in up to 97% of assayed children and adults (Addonizio et al., 1987). Tachycardia has been reported to occur in more than 75% of cases, a rate similar to that of altered mental state (Steingard et al., 1992). Autonomic disturbances such as blood pressure lability, tachypnea, and diaphoresis are reported less consistently. A case of NMS usually begins with rigidity, which may suggest a connection with EPS, but rigidity is a frequent EPS reaction to neuroleptics, and a specific association with NMS may be spurious. Addonizio et al. (1987) reported that rigidity preceded the fever in 59% of cases and simultaneous onset of rigidity and fever occurred in an additional 23% of cases. The onset of NMS is often within 2 weeks of neuroleptic initiation (Addonizio et al., 1987), and occasionally after a single dose (Klein et al., 1985). NMS has also occurred after small doses of neuroleptics used for nausea (Brower et al., 1989; Brown et al., 1991). Neither neuroleptic dose nor blood levels have been shown to be causative, and a general mechanism, the rate of increase in neuroleptic blood levels, has been suggested as provocative of NMS (Peterson et al., 1995). This is consistent with the observation that cases occur early in the course of neuroleptic treatment but seem less frequent when steady-state neuroleptic blood levels have been established. Some authors have suggested that high-potency neuroleptics present a greater risk for the development of NMS (Peterson et al., 1995; Susman & Addonizio, 1988). NMS has a variable course and outcome. Mortality rates in the range of 20% to 22% have been reported (Caroff, 1980; Shalev & Munitz, 1986); in addition to significant mortality, survival of an episode may be associated with persistent physical abnormalities, including renal, hepatic, or neuromuscular impairments (Addonizio et al., 1987; Pearlman, 1986; Steingard et al., 1992). Finally, there is wide variation in the time required for recovery to a premorbid level of functioning. NMS may have a different course in children. Most of the research has been done with adults, and one of the purposes of this review is to bring together what is known about childhood and adolescent forms in order to renew the attention of child psychiatrists to this condition. Previous reviews have used univariate statistical approaches to examine the efficacy of treatments and outcomes. Peterson et al. (1995) suggested that the “safety and efficacy of these medications for treating youth with NMS remain unclear” (p. 147). Also, Steingard et al. (1992) could not identify common factors associated with cases
Neuroleptic malignant syndrome in children and adolescents
437
culminating in death in children or adolescents. The optimal approach would be to establish an experimental design to answer these questions. During a lifethreatening illness, however, it is not feasible to experimentally restrict interventions. Alternatively, it is possible to explore these complexities by relying on methods that hold certain critical variables statistically fixed. This review will explore multiple regression approaches to identify signs associated with outcome and to synthesize information to match the presentation with the course and optimal treatment. Finally, this report will articulate the relevant information that should be included in future case reports of NMS and highlight evidence that prompt and strategic interventions can reduce the mortality and morbidity of this iatrogenic condition.
METHOD A preliminary review identified variables that should be included—variables germane to the diagnosis of NMS or previously associated with course or outcome. Thirty-eight variables were identified, including the following: demographic—age, gender, race; clinical—psychiatric diagnosis, medical condition, mental retardation; medication—all medications patient received, route and dose of medication, time from initiation of neuroleptics to beginning of NMS; NMS variables—the presence and extent of physical findings, laboratory abnormalities, elapsed time from the first symptom of NMS to its full presentation, time lapse from development of NMS until neuroleptics were discontinued, time from medication discontinuation to the resolution of NMS, and a list of all treatments and interventions. An extensive review of the literature via computerized literature searches (Medline, 1966–1998, PsychInfo, 1984–1998) was conducted. Articles were reviewed and all reference sections were checked for additional cases. All reports of cases aged 18 and younger were reviewed, independently, by two boardcertified child and adolescent psychiatrists, listing the findings for each of the 38 variables. A consensus meeting was held with a third author where discrepancies were resolved after reviewing the original articles. Three cases were excluded because a consensus diagnosis could not be achieved. Double data entry was used to minimize input errors. Statistical software used to run these analyses was Systat Version 5.2.1. Most of the reports were not complete with regard to the items that we planned to examine. Exclusion of incomplete reports would have eliminated a large percentage of the reports, and we elected to include as much information as possible. Because uniform reporting was not the rule, we provide sample sizes (N) for each variable. To evaluate the relative effectiveness of different treatments it was important to create, on a post hoc basis, a measure of severity of the NMS illness. A composite severity score based on the number of NMS signs reported for each case (see Table 24.1 for the 8 most frequent signs) plus the number of abnormal laboratory findings (see Table 24.2 for the eight most common measures) was constructed. These scores are reported in Table 24.3 as the NMS severity score. The statistical analysis proceeded from descriptive statistics of clinical signs (Table 24.1) and abnormal laboratory values (Table 24.2) and of the various outcome measures.
Annual progress in child psychiatry and child development 2000–2001
438
To examine the relations among overlapping clinical signs, which often occurred in clusters, some exploratory hierarchical multiple regression analyses (MRAs) are reported. A similar approach was taken to evaluate the relative efficacy of the multiple treatments that were administered simultaneously. MRAs were performed on variables if there were at least 10 df per predictor. Subjects The review identified 77 cases (49 males, 27 females, one gender unknown) reported in 61 articles (marked by an asterisk in the reference section). Ages ranged from .9 to 18 years (mean 14.8 years±3.96). Racial composition was as follows: Asian (15.6%), white (11.7%), Hispanic (6.5%%), African-American (3%), other (2.6%), and unlisted (59.8%). The psychiatric diagnoses for which these patients received pharmacotherapy included schizophrenia (24.3%), schizoaffective disorders (5.4%), bipolar disorder (17.6%), and other psychotic diagnosis (23.0%); in 17.6% of cases no psychiatric diagnosis was indicated. Eight patients were described as mentally retarded, another had microcephaly. Premorbid physical health was reported in 62 cases, 38.7% of whom had a medical problem (the most frequent being neurological, N=8).
TABLE 24.1 Clinical Presentation of Neuroleptic Malignant Syndrome (NMS)
Increased Blood Heart Pressure Fever Rate FluctuationRigidityTremorDystoniaDiaphoresisIncontin 47 37 58 19 23 23 9 Frequencya 56 b 62 60 57 62 52 54 54 53 Total N Percent 90.3 78.3 64.9 93.5 36.5 42.6 42.5 16. Sequelae 14 10 10 12 5 5 5 1 (N) Deaths (N) 6 5 2 6 2 2 3 1 Duration of NMSc Median 13.5 14 12.5 12 14 14 12 17 Mean 19.6 18.6 15.1 8 22.3 21.1 20.4 18. (SD) (20.8) (16.2) (14.2) (20.7) (26.3) (27) (26.5) (13. aNumber of cases in which the symptom was present. bNumber of cases in which the symptom was noted as present or absent. cDuration in days.
Neuroleptic malignant syndrome in children and adolescents
439
TABLE 24.2 Results of Laboratory Tests in Neuroleptic Malignant Syndrome (NMS)
Increased Increased Serum Liver CT Brain WBC CPK Sodium Enzymes EEG Scan Scan CSF 36 45 4 19 13 3 1 2 Frequencya b 48 48 18 30 28 18 6 27 Total N Percent 75.0 93.8 22.2 63.3 46.4 16.7 16.7 7.4 Sequelae (N) 7 11 1 5 9 2 0 2 Deaths (N) 2 1 2 0 4 0 0 0 Duration of NMSc Median 14 14 13 14 12 21 14 62.5 Mean 18.5 21 16 27 18.5 27.7 14 62.5 (SD) (6.8) (22) (8.7) (30) (18.3) (25.7) (79.9) Note. CPK=creatine phosphokinase; CSF=cerebrospinal fluid; CT=computed tomography; EEG=electroencephalograph; WBC=white blood cell count. aNumber of cases in which the test was abnormal. bNumber of cases in which the test was performed. cDuration in days.
RESULTS Pre-NMS Medications The number of different medications that children were receiving at the onset of NMS, including agents other than neuroleptics, ranged from 1 to 8 (mean 2.9±1.76; N=67). In 61.8% (34/55) of the cases, more than one neu-roleptic was administered concurrently. Approximately 72% received a highpotency neuroleptic (including 5.5% who received three high-potency agents), whereas 59.5% received up to two different low-potency neuroleptics. Of 65 patients, 17 received an anticholinergic agent, nine received lithium, nine received an anxiolytic, five were receiving an antidepressant, and seven were treated with one or more anticonvulsants. The duration of treatment with the offending neuroleptic agent(s) prior to the onset of NMS ranged from 2.5 hours to 168 days (mean 15.5 ± 32.29 days; N=64). Clinical Presentation To begin to associate characteristics of the clinical presentation of NMS with interventions and outcomes, three tables were constructed. Table 1 summarizes the association between presentation of NMS and outcome. Between two and eight NMS signs were mentioned in the reports for each case. In the table, it can be seen that fever
Annual progress in child psychiatry and child development 2000–2001
440
was listed in 56 of the 62 cases in which a patient’s temperature was mentioned and rigidity was equally prevalent. In addition, but not included in Table 1, an alteration in consciousness was seen in 72% (44/61), which included coma (19.4% or 11/61). Course Time from the appearance of the first symptom of NMS until the full syndrome developed varied from immediately to 18 days (mean 2.8 ±3.91 days; N=53). Time from the onset of NMS symptoms to the time of medication discontinuation ranged from immediately to 59 days (mean 4.4 ± 9.51 days; N =44); in 33 cases either neuroleptics were not discontinued or their discontinuation was not clearly reported. The time from medication discontinuation to the resolution of the NMS varied from immediately to 61 days (mean 11.8 ± 11.49; N=42). The total duration of NMS ranged from 1 to 119 days (mean 17.9±19.97; N=62). Outcome Outcome was clearly delineated in 65 patients; seven patients died and 15 cases resolved with physical sequelae, including residual rigidity (most commonly reported), brachial plexus palsy, residual dysarthria, liver function abnormalities, atelectasis, increased prolactin levels, and development of other abnormal movements. In Table 24.1 it can be seen that fever was associated with poor outcome; six of the seven children who died and 14 of the 15 with physical sequelae were reported to have fever. The relations between abnormal laboratory results and outcome are presented in Table 24.2. In addition, death was noted in four of the 13 cases with abnormal EEG findings, and the other nine cases suffered persistent sequelae. It appears that EEGs were performed primarily for patients with coma (11/13), which may explain the poor outcome associated with an abnormal EEG. Of the 31 cases for whom confusion was reported, 13 either had a sequela or died. Of the 11 with coma, five experienced a similar outcome (four of whom died). Thus, in the cases reviewed, 18 of the 22 patients with the most severe outcomes presented with an altered level of consciousness. To generate hypotheses for future work, we examined the characteristics of the seven cases with fatal outcomes. These children were somewhat younger (mean 12.5 versus 14.8 years), but their most striking characteristic was that the reports describing these cases were published earlier than the average report. Reports of a fatal outcome clustered in years prior to 1976, and no fatalities have been reported since 1986. The duration of illness in patients with fatal outcomes was substantially shorter (mean 7.7 days ±6.14) than in those who recovered without incident (mean 15.0 ± 14.58 days). It appears that the period of greatest mortal risk is early in the NMS course. The mean duration of illness for patients with physical sequelae was 29.8 days±29.93. Cases with sequelae took significantly longer (p=.007) to recover from the episode of NMS. Duration of illness provides a useful measure of severity.
Neuroleptic malignant syndrome in children and adolescents
441
Treatment The number of different interventions and medications per case ranged from one to 12, with an average of 3.6. Details concerning treatment interventions are summarized in Table 24.3. Among other things, it can be seen that the mean severity score for patients who received electroconvulsive therapy (ECT) was less than average, whereas children who received dantrolene and/or bromocriptine had significantly more signs and symptoms (t-test, t=5.15, p= .001) than the other children (mean 6.8±2.20 symptoms). Table 3, which includes rates of death and sequelae, shows that dantrolene and L-dopa were each associated with a death, wheres the other three agents were not. The rate of development of sequelae was lowest for anticholinergic agents (17%) and highest for ECT (44%).
TABLE 24.3 Treatment Described in the Case Reports of Neuroleptic Malignant Syndrome (NMS)
SupportiveNeurolepticsAnticholinergics/ LTreatment DC’D Amantadine BromocriptineDantrolene dop 35 50 17 18 19 8 Frequencya b 55 55 48 57 58 58 Total N Percent 63.6 90.9 35.4 31.6 32.8 13. Sequelae 7 15 3 7 5 3 (N) Deaths (N) 2 3 0 0 1 1 Duration of NMSc Median 12 12.5 14 13 15 32 Mean 14.9 19.2 19.9 25.7 21.3 35 (SD) (14.8) (21.6) (29.3) (29.5) (17.9) (20. NMS severity score Median 8 7 7.5 7 7 7 Mean 7.2 6.8 7.1 7.6 7.6 7.4 (SD) (2.1) (2.1) (2.5) (1.3) (1.4) (2.1 Note. DC’D=discontinued; ECT=electroconvulsive therapy. aNumber of reports in which the treatment was administered. bNumber of reports in which the treatment was mentioned. cDuration in days. Multiple Regression Analysis We next examined the relations between outcome and the demographic variables (age,
Annual progress in child psychiatry and child development 2000–2001
442
ethnicity, and gender). This information could help determine whether these factors should be controlled in further analyses. The demographic variables were unrelated to outcome. With regard to age, in the sample of 77 patients, four subjects were below the age of 6 years, six were between the ages of 6 and 10 years, 22 were between the ages of 11 and 15 years, and 45 patients were between 16 and 18 years old; the increased incidence with age probably reflects the lower rate of neuroleptic exposure in younger children. The first MRA examined outcome in relation to the use of high- and/or lowpotency neuroleptics. Previous reports, mainly involving adults, have implicated high-potency agents as presenting greater risk for NMS. Information on the base rate of neuroleptic exposure of children is not available, but in this survey low-potency agents were associated with a poorer outcome (β=.46, N =62, p=.01). We examined the triad of rigidity, fever, and tachycardia in the group of children who survived the episode (N=53) using duration of illness as the dependent measure (R2=.47, p<.001). Fever was related to longer duration of illness (β=.58, p=.03), whereas neither tachycardia nor rigidity was related to outcome. To examine the association between blood pressure fluctuations and outcome, we performed another hierarchical MRA. In this model (R2= .54, p<.001), fever was associated with poor outcome (β=1.04, p<.001) and blood pressure fluctuations were associated with briefer duration of illness (β =−.48, p=.01). This suggests that blood pressure fluctuations may reflect the influence of compensatory cardiovascular mechanisms. Another hierarchical MRA was done to examine the association between the different treatments and duration of the NMS episode. The overall model was highly significant (R2=0.46, p<.001). Many subjects received several medications, and this analysis was designed to separate the unique variance associated with each treatment. Only 53 subjects were available (cases with a fatal outcome were eliminated to permit us to use the duration parameter). The overall model was highly significant; dantrolene and ECT, which have been reported to be effective in adults (Addonizio et al., 1987), were not useful in this sample of children and adolescents. Anticholinergics (β=.27), bromocriptine (β=.33), and L-dopa (b=.31) were all significantly associated with shorter duration of illness.
DISCUSSION With a few exceptions NMS presents with a form and course in children that is similar to that found in adults. The triad of fever, tachycardia, and rigidity characterizes the presentation in both children and adults. NMS occurred almost twice as frequently in boys. This difference probably reflects the higher base rate of exposure to medication in boys and not a special vulnerability of males to NMS. Although the presentation of these children was similar to that in the broader age range studied by Pearlman (1986) and by Addonizio et al. (1987) in most regards, the children showed a higher rate of dystonia (40.8% versus 29% for all ages) and a lower rate of tremor (32.7% versus 48% for all ages). These differences probably reflect age differences in the rates of these extrapyramidal manifestations; dystonia is more common and tremor is less common in
Neuroleptic malignant syndrome in children and adolescents
443
these younger patients. Most importantly, the number of fatal outcomes has dropped sharply for all age groups. Pearlman noted in 1986 that mortality rates had decreased from 22% to 4% over the prior decade. Similarly, we have been unable to locate a report of a fatal outcome in a child or adolescent due to NMS that was published after 1986, whereas prior to that time the mortality rate was 21%. It seems likely that clinicians have become more aware of the dangers of NMS, and some combination of earlier detection, prompt neuroleptic discontinuation, and more aggressive treatment has moderated the fatal course of this iatrogenic disorder. An observation supporting this view is that there are early reports of five children for whom neuroleptics were continued after the onset of NMS, four of whom died. Although the mortality of NMS episodes has decreased with time, the rate at which cases resolve with sequelae has, if anything, increased. Fourteen of the 15 cases with residual physical abnormalities have been reported since 1983, suggesting that survivors of cases that previously might have been fatal are vulnerable to serious organ system damage. The conservative strategy for the management of anticholinergics remains confusing. Children receiving adjunctive anticholinergic therapy prior to the development of NMS experienced better outcomes if the anticholinergics were discontinued. For children not receiving anticholinergics prior to the development of NMS, initiation of anticholinergics was one of the more effective treatments (see Table 24.3). It is puzzling that the action of anticholinergics appears to depend on their temporal relationship to the development of NMS. This observation could benefit from further investigation. Polypharmacy appears to be prevalent in children and adolescents who develop NMS. Despite the fact that many of the children received a number of agents, those receiving low-potency drugs had a poorer outcome. In the group of 15 who developed sequelae, 13 were receiving at least one low-potency agent; similarly, five of the seven children who died were receiving a lowpotency agent. This observation may not contradict the suggestion of Pearlman (1986) and Addonizio et al. (1987) that high-potency agents are more likely to provoke NMS because information on the base rates of exposure to these agents is not available. Because of the seriousness of this condition, it seemed desirable to examine the data closely for clues about underlying mechanisms and optimal interventions. It would be more satisfying if there were a good animal model to pursue some of these leads. A number of indirect observations suggest that NMS is an aspect of the extrapyramidal actions of neuroleptics and reflects processes in the basal ganglia that are set in place by a relative decrease in the access of dopamine to the postsynaptic receptor. Precipitation of NMS does not appear to be related to the dose of neuroleptic but rather follows dose increase and can occur within hours of the first dose or, apparently, after months of maintenance treatment if there is an increase in dose. Of interest is that a syndrome similar to NMS can occur in patients with Parkinson disease if dopaminergic agents are stopped (Friedman et al., 1985; Sechi et al., 1984; Simpson & Davis, 1984). Both the initiation of neuroleptics and the discontinuation of antiparkinson agents are associated with a decrease in the access of dopamine at the postsynaptic receptor. A viable animal model would permit exploration of this observation. It seems possible that the increased muscle activity associated with the rigidity
Annual progress in child psychiatry and child development 2000–2001
444
generates metabolic heat. Consistent with this, increased CPK appeared at a high rate where it was assayed, and elevated CPK was associated with a high rate of sequelae. In the MRA of the triad of NMS signs, when the contributions of fever and tachycardia were controlled, the contribution of rigidity was not related to duration of illness, suggesting that its role in determining outcome may be mediated by its efficiency in provoking fever; it is not clear whether some brainstem disturbance in heat regulation also contributes to the development of NMS because rigidity is common and NMS is rare. The MRAs, which found a positive role for blood pressure fluctuations, suggested that an ability to adapt to cardiovascular stress is positively related to recovery from NMS. The information from the articles we reviewed is limited in obvious and subtle ways. Few of the articles contained uniform information. It might be helpful if the major journals agreed on editorial policies that were more prescriptive as to the material that should be included in case reports. We suggest that the information included in our Results section be included in case reports and that the presence of these items be confirmed or denied. Another issue that needs to be acknowledged is that it is unknown whether published cases of NMS are representative. In the absence of an animal model, it is unlikely that experimental designs can be applied to the study of NMS. These difficulties increase our dependence on statistical methods to isolate traits that contribute to the course and outcome of NMS. Where we may have violated some of the assumptions of our statistical models, we have been conservative in our clinical recommendations. Until a prospective multisite effort can provide more interpretable results, it seems justified to exploit fragmentary data as fully as possible. Clinical Implications In addition to the early identification of the signs and symptoms of NMS, followed by the prompt discontinuation of neuroleptics and management of the anticholinergic status, it is important to provide the medically necessary supportive measures (e.g., fever reduction and intravenous hydration). Our results suggest that bromocriptine be considered a firstline treatment in this condition. As Table 24.3 suggests, the more favorable outcome with bromocriptine is not an artifact of treating less afflicted children. However, the use of Ldopa, despite being related to shorter illness duration, is also associated with a 50% rate of morbid outcomes. Anticholinergics (initiated after the development of NMS), bromocriptine, and ECT were not associated with a death. It should be remembered that fulminating cases likely elicited heroic efforts that involved multiple treatments, and the poor outcomes of several of the medications listed may result from their use in crisis situations.
REFERENCES1 Addonizio G (1986). Neuroleptic malignant syndrome and ECT. Proceedings of the Annual Meeting of the American Psychiatric Association, p 172. Cited in: Pearlman CA (1986), Neuroleptic malignant syndrome: a review of the literature. J Clin
Neuroleptic malignant syndrome in children and adolescents
445
Psychopharmacol 6:257–273.* Addonizio G, Susman V, Roth S (1987). Neuroleptic malignant syndrome: review and analysis of 115 cases. Biol Psychiatry 22:1004–1020. Addonizio G, Susman VL (1987), ECT as a treatment alternative for patients with symptoms of neuroleptic malignant syndrome: clinical reports. J Clin Psychiatry 48:102–105.* American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition. Washington, DC: Author. Araki M, Takagi A, Higuchi I, Sugita H (1988). Neuroleptic malignant syndrome: caffeine contracture of single muscle fibers and muscle pathology. Neurology 38:297– 301.* Bhatia MS, Singhal PK, Bohra N, Kaur N (1988). Neuroleptic malignant syndrome. Indian Pediatr 25:788–790.* Brower RD, Dreyer CF, Kent TA (1989). Neuroleptic malignant syndrome in a child treated with metoclopramide for chemotherapy-related nausea. J Child Neurol 4:230– 232.* Brown FE, Nierenberg DW, Nordgren RE, Taylor RM, Rozycki AA (1991). Neuroleptic malignant syndrome: occurrence in a child after reconstructive surgery. Plast Reconstr Surg 87:961–964.* Garden PA, Mitchell SL (1988). Acute psychosis plus neuroleptic malignant syndrome as complications of therapy in an adolescent with acute lymphoblastic leukaemia. Aust J Hosp Pharm 18:205–207.* Caroff S (1980). The neuroleptic malignant syndrome. J Clin Psychiatry 41:79–83. Corless JD, Buchanan DS (1965). Phenothiazine intoxication in children: a report of three cases. JAMA 194:565–567.* Dave M (1995). Two cases of risperidone-induced neuroleptic malignant syndrome. Am J Psychiatry 152:1233–1234. Delay J, Deniker P (1968). Drug-induced extrapyramidal syndromes. In: Handbook of Clinical Neurology: Diseases of the Basal Ganglia, Vol 6. Vinken PJ, Bruyn GW, eds. New York: Elsevier/North Holland, pp 248–266. 1
References followed by an asterisk were the source of NMS cases. Many of these references are not cited individually in the text
de Rohan Chabot P, Elkharrat D, Conso F, Bismuth C, Goulon M (1982). Syndrome malin des neuroleptiques action benefique du dantrolene sur l’hyperthermie et la rigidite musculaire. Nouv Presse Med 11:1067–1069.* Dhib-Jalbut S, Hesselbrock R, Mouradian MM, Means ED (1987). Bromocriptine treatment of neuroleptic malignant syndrome. J Clin Psychiatry 48:69–73.* Diamond JM, Hayes DD (1986). A case of neuroleptic malignant syndrome in a mentally retarded adolescent. J Adolesc Health Care 7:419–422.* Friedman JH, Feinberg SS, Feldman RG (1985). A neuroleptic malignant like syndrome due to levodopa therapy withdrawal. JAMA 254:2792–2795. Gabris G, Muller C (1983). La catatonie dite “precieuse.” Encephale 9:365–385.* Geller B, Greydanus DE (1979). Haloperiodol-induced comatose state with hyperthermia and rigidity in adolescence: two case reports with a literature review. J Clin Psychiatry
Annual progress in child psychiatry and child development 2000–2001
446
40:102–103.* Goulon M, Rohan-Chabot P, Elkharrat D, Gajdos P, Bismuth C, Conso F (1983). Beneficial effects of dantrolene in the treatment of neuroleptic malignant syndrome: a report of two cases. Neurology 33:516–518.* Guerrero RM, Shifrar KA (1988). Diagnosis and treatment of neuroleptic malignant syndrome. Clin Pharm 7:697–701.* Harsch HH (1987). Neuroleptic malignant syndrome: physiological and laboratory findings in a series of nine cases. J Clin Psychiatry 48:328–333.* Hayashi K, Chichara E, Sawa T, Tanaka Y (1993). Clinical features of neuroleptic malignant syndrome in basal ganglia disease . Anaesthesia 48:499–502.* Imagi M (1977). Neuroleptic malignant syndrome: skeletal muscle findings [in Japanese]. Rinsho Shinkeigaku 7:976. Cited in: Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Itoh H, Ohtsuka N, Ogita K, Yagi G, Miura S, Koga Y (1977). Malignant neuroleptic syndrome: its present status in Japan and clinical problems. Folia Psychiatr Neurol Jpn 31:565–576.* Joshi PT, Capozzoli JA, Coyle JT (1991). Neuroleptic malignant syndrome: lifethreatening complication of neuroleptic treatment in adolescents with affective disorder. Pediatrics 87:235–239.* Keck PE, Pope HG, Cohen BM, McElroy SL, Nierenberg AA (1989). Risk factors for neuroleptic malignant syndrome: a case-control study. Arch Gen Psychiatry 46:914– 918.* Keck PE, Pope HG, McElroy SL (1991). Declining frequency of neuroleptic malignant syndrome in a hospital population. Am J Psychiatry 148:880–882.* Klein SK, Levinsohn MW, Blumer JL (1985). Accidental chlorpromazine ingestion as a cause of neuroleptic malignant syndrome in children. J Pediatr 107:970–973.* Kontaxakis V, Stefanis C, Markidis M, Tserpe V (1988). Neuroleptic malignant syndrome in a patient with Wilson’s disease. J Neurol Neurosurg Psychiatry 51:1001– 1002.* Koponen H, Repo E, Lepold U (1988). Neuroleptic malignant syndrome. Biol Psychiatry 24:943–945.* Kubota T (1993). Neuroleptic malignant syndrome induced by nemonapride. Acta Neurol (Napoli) 15:142–144.* Latz SR, McCracken JT (1992). Neuroleptic malignant syndrome in children and adolescents: two case reports and a warning. J Child Adolesc Psychopharmacol 2:123– 129.* Levenson JL (1985). Neuroleptic malignant syndrome. Am J Psychiatry 142:1137–1145. Magen JG, D’Mello D (1995). Acute lymphocytic leukemia and psychosis: treatment with electroconvulsive therapy. Ann Clin Psychiatry 7:133–137.* Mancias P, Kramer L, Butler IJ (1995). Amoxapine overdose in a young man: a transient mitochondrial abnormality? Pharmacotherapy 15:528–532.* Merry SN, Werry JS, Merry AF, Birchall N (1986). The neuroleptic malignant syndrome in an adolescent. J Am Acad Child Psychiatry 25:284–286.* Moore N, O’Donohoe NV, Monaghan H (1986). Neuroleptic malignant syndrome. Arch Dis Child 61:793–794.*
Neuroleptic malignant syndrome in children and adolescents
447
Moyes DG (1973). Malignant hyperpyrexia caused by trimeprazine. Br J Anaesth 45:1163–1164.* Nolen WA, Zwaan WA (1990). Treatment of lethal catatonia with electroconvulsive therapy and dantrolene sodium: a case report. Acta Psychiatr Scand 82:90–92.* Ogita K (1980), Syndrome malin. Seishin Igaku 22:1229–1237. Cited in: Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Otsuka K, Noga Y, Saito S et al. (1974). Syndrome malin due to neuroletica. Jpn J Clin Psychiatry 3:961–973. Cited in: Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Padgett R, Lipman E (1989). Use of neuroleptics after an episode of neuroleptic malignant syndrome. Can J Psychiatry 34:323–325.* Parke TJ, Wheatley SA (1992). Anaesthesia in the neuroleptic malignant syndrome. Anaesthesia 47:908–909.* Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273. Perez de los Cobos JC, O’Neil A, Lopez-Ibor JJ Jr (1992). Galactorrhea in the neuroleptic malignant syndrome. Eur Neuropsychopharmacol 2:87–89.* Peterson SE, Myers KM, McClellan J, Crow S (1995). Neuroleptic malignant syndrome: three adolescents with complicated courses. J Child Adolesc Psychopharmacol 5:139– 149.* Pope HG, Aizley HG, Keck PE, McElroy SL (1991). Neuroleptic malignant syndrome: long-term follow-up of 20 cases. J Clin Psychiatry 52:208–212.* Pope HG, Keck PE, McElroy SL (1986). Frequency and presentation of neuroleptic malignant syndrome in a large psychiatric hospital. Am J Psychiatry 143:1227–1233.* Preston J (1959). Central nervous system reactions to small doses of tranquilizers. Am Pract 10:627–630.* Revuelta E, Border R, Piquet T, Ghawche F, Destee A, Goudemand M (1994). Catatonie aigue et syndrome malin des neuroleptiques. Encephale 20:351–354.* Ries RK, Schuckit MA (1980). Catatonia and autonomic hyperactivity. Psychosomatics 21:349–350.* Rojtman M, Apter E, Lahav S, Tiano S (1981). Amantadine in malignant neutoleptic syndrome [in Hebrew]. Harefuah 100:333. Cited in: Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Rosebush P, Stewart T (1989). A prospective analysis of 24 episodes of neuroleptic malignant syndrome. Am J Psychiatry 146:717–725.* Rosebush PI, MacQuee GM, Clarke JTR, Callahan JW, Strasberg PM, Mazurek MF (1995). Late-onset Tay-Sachs disease presenting as catatonie schizophrenia: diagnostic and treatment issues. J Clin Psychiatry 56:347–353.* Sechi G, Tanda F, Mutani R (1984). Fatal hyperpyrexia after withdrawal of levodopa. Neurology 34:249–251. Shalev A, Munitz R (1986). The neuroleptic malignant syndrome: agent and host interaction. Acta Psychiatr Scand 73:337–347. Shaw A, Mathews EE (1995). Postoperative neuroleptic malignant syndrome. Anaesthesia 50:246–247.*
Annual progress in child psychiatry and child development 2000–2001
448
Silva RR, Munoz DM, Perlmutter IR, Woo C, Chouchani S (1994). Neuroleptic malignant syndrome in children and adolescents. In: Scientific Proceedings, American Academy of Child and Adolescent Psychiatry, 40th Annual Meeting, Vol X, New York, pp 48–49.* Simpson DM, Davis GC (1984). Case report of neuroleptic malignant syndrome associated with withdrawal from amantadine. Am J Psychiatry 141:796–797. Singer S, Richards C, Boland RJ (1995). Two cases of risperidone-induced neuroleptic malignant syndrome. Am J Psychiatry 152:1234. Slack T, Stoudemire A (1989). Reinstitution of neuroleptic treatment with molidone in a patient with a history of neuroleptic malignant syndrome. Gen Hosp Psychiatry 11:365–367.* Steingard R, Khan A, Gonzalez A, Herzog D (1992). Neuroleptic malignant syndrome: review of experience with children and adolescents. J Child Adolesc Psychopharmacol 2:183–198.* Sumiyoshi A, Sakaki T, Oguchi T et al. (1980). Four cases of syndrome malin induced by neuroleptics [in Japanese]. Masui to Sosei 16(suppl):83–87. Cited in: Pearlman CA (1986). Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Susman VL (1986). Managing an acutely manic 17-year-old girl with neuroleptic malignant syndrome. Hosp Community Psychiatry 37:771–788.* Susman VL, Addonizio G (1988). Recurrence of neutoleptic malignant syndrome. J Nerv Ment Dis 176:234–241.* Tenenbein M (1985–1986). The neuroleptic malignant syndrome: occurrence in a 15yearold boy and recovery with bromocriptine therapy. Pediatr Neurosci 12:161–164.* Thacker AK, Radhakrishnan K, Maloo JC, Bohlaga NH (1990). Neuroleptic malignant syndrome in a girl without psychosis. Br J Clin Pract 44:425–427.* Thery P (1979). Contribution a l’etude clinique et anatomopathologique du syndrome malin des neuroleptiques. Medical thesis, Bordeaux. Cited in: Pearlman CA (1986), Neuroleptic malignant syndrome: a review of the literature. J Clin Psychopharmacol 6:257–273.* Thomas K, Rajeev KK, Abraham OC, Archana S, Cherian AM (1992). Management of neuroleptic malignant syndrome: a series of eight cases. J Assoc Physicians India 41:91–93.* Verhoeven WMA, Elderson A, Westenberg HGM (1985). Neuroleptic malignant syndrome: successful treatment with bromocriptine. Biol Psychiatry 20:680–684.* Walker WD (1988). Spectrum concept of neuroleptic malignant syndrome (letter). Br J Psychiatry 153:574.* Woodbury MM, Woodbury MA (1992). Neuroleptic-induced catatonia as a stage in the progression toward neuroleptic malignant syndrome. J Am Acad Child Adolesc Psychiatry 31:1161–1164.*
PART V: TREATMENT ISSUES
25 AHA Scientific Statement: Cardiovascular Monitoring of Children and Adolescents Receiving Psychotropic Drugs Howard Gutgesell, Dianne Atkins, Robyn Barst, Marcia Buck, Wayne Franklin, Richard Humes, Richard Ringel, Robert Shaddy, and Kathryn A.Taubert
Reports of sudden deaths of children and adolescents treated with psychotropic medications have raised concerns regarding the appropriateness of this therapy, as well as the advisability of baseline and periodic electrocardiographic (ECG) monitoring of such patients (Biederman et al, 1989, 1995; Popper & Zimnitzky, 1995; Werry et al., 1995). What follows is a review of the drug effects on the ECG, cardiovascular effects of the commonly used psychotropic medications in children and adolescents, a summary of potentially dangerous drug interactions, and recommendations for cardiovascular monitoring.
POTENTIAL MECHANISMS FOR SUDDEN DEATH Although medications can potentially cause sudden, unexpected death by a variety of mechanisms (e.g., seizures, central nervous system depression, or coronary artery spasm), cardiac arrhythmias are the most frequent cause. In particular, a unique form of ventricular tachycardia termed torsade de pointes has been recognized as the arrhythmia responsible for the so-called proarrhythmic effect of several antiarrhythmia drugs, and recent evidence has pointed to a similar mechanism in syncope and deaths related to other medications (Roden et al., 1986) and in the familial long-QT syndromes (Keating & Sanguinetti, 1996). The common feature of these conditions is delayed repolarization of the myocardium (related to abnormal sodium or potassium currents) with resultant prolongation of the QT interval of the ECG. This appears to leave the myocardium vulnerable to ventricular tachycardia, primarily in the setting of bradycardia but occasionally in association with exercise. Other ECG abnormalities, such as sinus node depression, second- or thirddegree heart block, and supraventricular tachycardia, seem unlikely causes of sudden death in patients receiving psychotropic medications. Additionally, the rSR’ pattern in lead V1, sometimes referred to as incomplete right bundle-branch block or right ventricular conduction delay,
AHA scientific atatement
451
is a normal childhood variant and is not a risk factor for sudden death.
SPECIFIC DRUGS The major cardiovascular and electrophysiological effects of the commonly used psychotropic drugs are listed in Table 25.1. Stimulants such as the amphetamines and methylphenidate (Ritalin) cause slight but clinically insignificant increases in heart rate and blood pressure. The tricyclic antidepressants (TCAs) imipramine and desipramine have been associated with at least seven reported deaths in young patients (Popper & Zimnitzky, 1995; Riddle et al., 1991). The precise mechanism of death has not been documented. Some of the patients have had toxic levels of drugs, and at least two had risk factors for sudden death (a coronary anomaly in one and a family history of sudden death in another). The ECG effects of TCA administration include an increase in heart rate (by 20% to 25%), prolongation of the PR interval (by 5% to 10%), increase in QRS duration (by 7% to 25%), and prolongation of the QT interval (by 3% to 10%). Although 8% of patients have a corrected QT interval (QTc) of >460 milliseconds, the mean value of the QTc remained within normal limits for pediatric patients (Biederman et al., 1989; Fletcher et al., 1993; Schroeder et al., 1989; Wilens et al., 1993, 1996). Malignant arrhythmias, in particular torsade de pointes, have not been documented except for the ventricular fibrillation observed in the emergency room in the patient with a family history of sudden death. The selective serotonin reuptake inhibitors have minimal cardiovascular effects; deaths have been rare, even with massive overdose. Clonidine, a widely used antihypertensive medication, has been associated with two deaths in patients who also received methylphenidate, but the mechanism for these deaths is unknown and may have been sudden cessation of treatment.
TABLE 25.1 Cardiac Effects of Psychotropic Medications
Class
Medication
Cardiac Effects and Comments
Stimulants
Methylphenidate (Ritalin)
Cardiovascular Mild tachycardia and increase in blood pressure Adrenergic blockers are inhibited Sympathomimetics are enhanced Report of sudden death in 1 child treated simultaneously with
Recommendations for Cardiovascular Monitoring No specific cardiovascular monitoring is indicated
Annual progress in child psychiatry and child development 2000–2001
452
clonidine Pemoline (Cylert) Minimal cardiac effects No specific Dextroamphetamine Hepatotoxicity cardiovascular Tachycardia/palpitations monitoring is indicated No specific cardiovascular monitoring is indicated Antidepressants Desipramine Cardiovascular 1. Baseline history TCAs (Norpramin); Rare reports of sudden and physical imipramine death 2. Current Prolongation of the QTc, medication history PR, QRS 3. Baseline ECG Sinus tachycardia measuring: Other PR=200 msec Cimetidine increases QRS duration=120 levels of TCA msec Ritalin decreases the QTc=460 msec metabolism 4. Follow-up ECG May increase action of and history after a other sympathomimetic steady state is agents achieved on 3–5 Watch drug-drug mg/kg for interactions! desipramine and 3– 5 mg/kg for imipramine Selective Fluoxetine (Prozac) Cardiovascular No specific serotonin No significant ECG cardiovascular reuptake change monitoring is inhibitors Interacts with protein- indicated bound medications such Digoxin and as digoxin and warfarin warfarin dose may need adjustment Sertraline (Zoloft) Cardiovascular No significant ECG abnormalities Mild tachycardia Other No MAOI Paroxetine (Paxil) No significant cardiac Possible interaction effects with P450 system P450: evaluate other medications
AHA scientific atatement
Other Bupropion antidepressants (Wellbutrin) Lithium
Other psychotropic agents a2Adrenergic agonist
Antipsychotics/ neuroleptics Phenothiazines
Butyrophenones
Clonidine (Catapres)
453
Little information on cardiac effects Dopamine agonist Cardiovascular Reports of arrhythmias Flattened T Waves Other Monitor levels Diuretics decrease renal clearance (especially thiazide diuretics) Cardiovascular Hypotension Rebound hypertension when discontinuing
No specific cardiovascular monitoring is indicated No specific cardiovascular monitoring is indicated
1. Monitor blood pressure when medication is started 2. Monitor blood pressure when weaning from the medication 3. No ECG monitoring is necessary
Guanfacine Similar to clonidine (Tenex) Chlorpromazine Cardiovascular 1. Baseline history Thioridazine Tachycardia and physical Mesoridazine (anticholinergic) 2. Current Perphenazine Hypotension medication history Trifluoperazine Prolongation of the 3. Baseline ECG Fluphenazine QTc (be aware of measuring: other medications PR=200 msec that prolong the QTc) QRS duration=120 msec QTc=460 msec 4. Follow-up ECG and history after therapeutic levels are reached Haloperidol Cardiovascular 1. Baseline history (Haldol) Tachycardia and physical (anticholinergic) 2. Current Hypotension medication history
Annual progress in child psychiatry and child development 2000–2001
454
Prolongation of the 3. Baseline ECG QTc (be aware of measuring: other medications PR=200 msec that prolong the QTc) QRS duration=120 msec QTc=460 msec 4. Follow-up ECG and history after therapeutic levels are reached Diphenylbutylpiperidine Pimozide Cardiovascular 1. Baseline history (Orap) Prolongation of the and physical QTc (be aware of 2. Current other medications medication history that prolong the QTc) 3. Baseline ECG measuring: PR=200 msec QRS duration=120 msec QTc=460 msec 4. Follow-up ECG and history after therapeutic levels are reached Dihydroindolone Molindone Sinus tachycardia No specific (Moban) One of the few cardiovascular tranquilizers without monitoring is a warning about the indicated QTc Note. ECG=electrocardiographic; MAOI=monoamine oxidase inhibitor; TCA=tricyclic anti depressant.
DRUG INTERACTIONS Many psychotropic medications are metabolized by the cytochrome P450 system, an enzyme system that may be inhibited by a multitude of medications (Slaughter & Edwards, 1995; Table 25.2). Adverse effects have occurred when the P450 system is inhibited, which leads to elevated levels of medications that prolong the QT interval and produce ventricular tachycardia (torsade de pointes). Most notable have been deaths related to torsade de pointes from nonsedating histamine-blocking agents such as terfenadine (Seldane) and astemizole (Hismanal)
AHA scientific atatement
455
(Woosley et al., 1993). Many of these episodes were associated with coadministration of other medications such as macrolide antibiotics or imidazole antifungal agents (Table 25.2). Other classes of medications that inhibit or are metabolized by the cytochrome P450 system include antidepressants, calcium channel blockers, histamine blockers,
TABLE 25.2 Potential Drug Interactions
Group A. Drugs metabolized by the cytochrome P450 enzyme systemsa 1. Psychotropics Clomipramine Amitriptyline Desipramine Imipramine Nortriptyline Thioridazine 2. Antiepileptics Carbamazepine Phenytoin Valproic acid 3. Antiarrhythmics Amiodarone Diltiazem Encainide Felodipine Flecainide Metoprolol Nefedipine Propranolol Quinidine Timolol Verapamil 4. Other Cisapride Cyclosporine Lovastatin Omeprazole Tacrolimus Terfenadine Warfarin Group B. Drugs inhibiting the cytochrome P450 enzyme systemsa 1. Antibiotics Erythromycin Clarithromycin Fluconazole Intraconazole Ketoconazole Miconazole 2. Psychotropics Fluoxetine Fluvoxamine Haloperidol Paroxetine 3. Other Cimetidine Grapefruit juice Quinidine aConcomitant use of drugs from group A and group B may result in augmentation of pharmacologic and toxic effects of group A drugs. Drug reduction and careful monitoring is required.
Annual progress in child psychiatry and child development 2000–2001
456
gastrointestinal motility agents, and steroids. Prolongation of the QT interval and torsade de pointes have been reported in young children taking cisapride. Antiarrhythmic drugs of class Ia (e.g., disopyramide, procainamide, and quinidine) and class III (e.g., amiodarone and sotalol) likewise prolong the QT interval; therefore, concomitant use of psychotropic medications with these drugs is not recommended.
FAMILY HISTORY In some families, syncope and sudden death have been related to familial prolongation of the QT interval and torsade de pointes. Such individuals are at increased risk for arrhythmias due to medications that prolong the QT interval. Drugs that prolong the QT interval are contraindicated in patients with familial long-QT syndrome.
RECOMMENDATIONS 1. Before therapy with psychotherapeutic agents is initiated, a careful history should be obtained, with special attention to symptoms such as palpitations, syncope, or near syncope. Medication use (prescribed and over-thecounter) should be determined. The family history should be reviewed with reference to the long-QT syndrome or other causes of sudden, unexplained death. Detection of these symptoms or risk factors warrants a cardiovascular evaluation by a pediatric cardiologist before initiation of therapy. 2. At follow-up visits, patients receiving psychotropic drug therapy should be questioned about the addition of any drugs and the occurrence of any of the preceeding symptoms. The physical examination should include determination of heart rate and blood pressure. 3. Despite the lack of understanding of the mechanism of sudden death in young patients, taking TCAs or demonstration that ECG monitoring could prevent these deaths, ECG monitoring at baseline and during chronic therapy has been recommended (Biederman et al., 1989; Elliott & Popper, 1990; Wagner & Fershtman, 1993). Until more data are available, it seems prudent to obtain an ECG at baseline before TCA or phenothiazine therapy is begun (primarily to detect unsuspected instances of long-QT syndrome) and another when steady state is achieved. If the sustained resting heart rate is >130 beats per minute, the PR interval is >200 milliseconds, QRS is >120 milliseconds, or QTc is >460 milliseconds, or if symptoms such as palpitations, near syncope, or syncope develop, alternative therapy may need to be considered along with pediatric cardiology consultation. 4. Concomitant use of psychotropic drugs and other drugs that are metabolized by or inhibit the P450 enzyme system should be avoided.
AHA scientific atatement
457
REFERENCES Biederman J, Baldessarini RJ, Wright V, Knee D, Harmatz JS, Goldblatt A (1989). A double-blind placebo controlled study of desipramine in the treatment of ADD, II: serum drug levels and cardiovascular findings . J Am Acad Child Adolesc Psychiatry 28:903–911. Biederman J, Thisted RA, Greenhill LL, Ryan ND (1995). Estimation of the association between desipramine and the risk for sudden death in 5- to 14-year-old children. J Clin Psychiatry 56:87–93. Elliott GR, Popper CW (1990). Tricyclic antidepressants: the QT interval and other cardiovascular parameters. J Child Adolesc Psychopharmacol 1:187–189. Fletcher SE, Case CL, Sallee FR, Hand LD, Gillette PC (1993). Prospective study of the electrocardiographic effects of imipramine in children. J Pediatr 122:652–654. Keating MT, Sanguinetti MC (1996). Molecular genetic insights into cardiovascular disease. Science 272:681–685. Popper CW, Zimnitzky B (1995). Sudden death putatively related to desipramine treatment in youth: a fifth case and a review of speculative mechanisms. J Child Adolesc Psychopharmacol 5:283–300. Riddle MA, Nelson JC, Kleinman CS, et al. (1991). Sudden death in children receiving Norpramin: a review of three reported cases and commentary. J Am Acad Child Adolesc Psychiatry 30:104–108. Roden DM, Thompson KA, Hoffman BF, Woosley RL (1986). Clinical features and basic mechanisms of quinidine-induced arrhythmias. J Am Coll Cardiol 8 (suppl A):73A–78A. Schroeder JS, Mullin AV, Elliott GR, et al. (1989). Cardiovascular effects of desipramine in children [see comments in J Am Acad Child Adolesc Psychiatry 1989; 28:964–965]. J Am Acad Child Adolesc Psychiatry 28:376–379. Slaughter RL, Edwards DJ (1995). Recent advances: the cytochrome P450 enzymes. Ann Pharmacother 29:619–624. Wagner KD, Fershtman M (1993). Potential mechanism of desipramine-related sudden death in children. Psychosomatics 34:80–83. Werry JS, Biederman J, Thisted R, Greenhill L, Ryan N (1995). Resolved: cardiac arrhythmias make desipramine an unacceptable choice in children. J Am Acad Child Adolesc Psychiatry 34:1239–1245; discussion 1245–1248. Wilens TE, Biederman J, Baldessarini RJ, et al. (1996). Cardiovascular effects of therapeutic doses of tricyclic antidepressants in children and adolescents. J Am Acad Child Adolesc Psychiatry 35:1491–1501.
Annual progress in child psychiatry and child development 2000–2001
458
Wilens TE, Biederman J, Baldessarini RJ, Puopolo PR, Flood JG (1993). Electrocardiographic effects of desipramine and 2-hydroxydesipramine in children, adolescents, and adults treated with desipramine. J Am Acad Child Adolesc Psychiatry 32:798–804. Woosley RL, Chen Y, Freiman JP, Gillis RA (1993). Mechanism of the cardiotoxic actions of terfenadine [see comments in JAMA 1993; 269:1550–1552]. JAMA 269:1532–1536.
Part VI SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
PART VI: SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
26 Twenty Years’ Research on Peer Victimization and Psychosocial Maladjustment: A MetaAnalytic Review of Cross-Sectional Studies David S.J.Hawker and Michael J.Boulton
Cross-sectional quantitative designs are often used to investigate whether peer victimization is positively related to psychosocial maladjustment. This paper presents a meta-analytic review of crosssectional studies, published between 1978 and 1997, of the association of peer victimization with psychosocial maladjustment. Mean effect sizes were calculated for the association between peer victimization and each form of maladjustment (depression, loneliness, generalized and social anxiety, and global and social selfworth) assessed. The results suggested that victimization is most strongly related to depression, and least strongly related to anxiety. There was no evidence that victimization is more strongly related to social than to psychological forms of maladjustment. Effect sizes were stronger when the same informants were used to assess both victimization and maladjustment than when different informants were used. There were some design limitations to the studies reviewed, but all together their results provide a strong background for more complex research into the course and treatment of victims’ distress. Key Words: Bullying, cross-cultural, internalizing disorder, meta-analysis, peer relationships, victimization.
INTRODUCTION Peer victimization is the experience among children of being a target of the aggressive behavior of other children, who are not siblings and not necessarily agemates. Peer victimization is a problem of growing concern for researchers, educators, and clinicians (e.g., Ambert, 1995; Dawkins, 1995; Hazler & Hoover, 1996; Olweus, 1993a; Ross, 1996; Slee & Rigby, 1994). Children targeted for peer aggression are variously described as being bullied (e.g., Olweus, 1993a; Rigby, 1996; Whitney & Smith, 1993), being victimized (e.g., Crick & Grotpeter, 1996; Perry, Kusel, & Perry, 1988), or sometimes as being rejected (e.g., Vernberg, 1990). In this paper, we refer to studies which used any of
Annual progress in child psychiatry and child development 2000–2001
462
these terms, provided that they made some measurement of the experience of being a target of peers’ aggressive behavior. In recent years growing numbers of investigators have asked whether victims of peer aggression experience psychosocial maladjustment (depression, anxiety, low self-esteem, and the like). It is clearly important to know the answer to this question, so that children’s distress does not go unrecognized. A common way of addressing it empirically is with cross-sectional designs. In these designs quantitative methods are used to investigate the relation between children’s experiences of peer victimization and maladjustment, both assessed at a single point in time. Cross-sectional studies of this type are increasing in number rapidly. At least six were published in 1998 alone (Craig, 1998; Crick & Bigbee, 1998; Graham & Juvonen, 1998; Kumpulainen et al., 1998; Salmon, James, & Smith, 1998; Stanley & Arora, 1998). Independently of such studies, converging theoretical and empirical perspectives suggest that victims may suffer greater psychosocial maladjustment than nonvictims. First, a number of theorists have argued that certain social psychological experiences, reminiscent of peer victimization, play a central role in the development of depression and of other forms of psychosocial maladjustment. For example, Gilbert (1992) outlined how attacks on peer-group rank, with strong similarities to physical victimization, maintain depression; Baumeister and Leary (1995) suggested that threats to social bonds (cf. relational victimization) can lead to anxiety, loneliness, or depression. Second, empirical research has shown that certain types of maladjustment (such as loneliness, depression, anxiety, and low self-esteem) are positively associated with such peer relationship difficulties as submissiveness, social withdrawal, and unpopularity with peers (e.g., Parkhurst & Asher, 1992; Strauss, Lahey, Frick, Frame, & Hynd, 1988; Vosk, Forehand, Parker, & Rickard, 1982; Walker & Greene, 1986). These peer relationship difficulties are themselves positively related to peer victimization (e.g., Björkqvist, Ekman, & Lagerspetz, 1982; Boivin, Hymel, & Bukowski, 1995; Boulton & Smith, 1994; Perry et al., 1988; Schwartz, Dodge, & Coie, 1993). Thus there are strong a priori reasons to hypothesize that the pattern of results, from cross-sectional studies of peer victimization and psychosocial maladjustment, will show that these two experiences are positively related. Unfortunately, there has been no systematic or meta-analytic review of crosssectional studies of this type. As a consequence there is little awareness in the literature about the strength of empirical evidence concerning victims’ distress. Investigators typically cite a small number of results, from a limited range of studies, which may not be representative. The absence of evaluative reviews is unfortunate because many of the forms of psychosocial maladjustment in-vestigated separately in these studies are conceptually and empirically related among themselves (see, e.g., Eason, Finch, Brasted, & Saylor, 1985; Leary, 1990; West, Kellner, & Moorewest, 1986). In this paper we address the question of victims’ psychosocial maladjustment not with a new cross-sectional study, but by collating and evaluating the results of previous studies in a metaanalysis. Meta-analysis offers a quantitative summary of the effect sizes reported in quantitative research papers, and reduces the bias inherent in purely qualitative review papers (e.g., Rosenthal, 1995).
Twenty Years’ research on peer victimization and psychosocial maladjustment
463
Forms of Victimization and Adjustment Cross-sectional studies have typically measured peer victimization by asking respondents how much they or others experience being the targets of particular examples of aggression, such as being harassed, hit, or teased. In some studies (e.g., O’Moore & Hillery, 1991) these experiences are combined in a single item as examples of “bullying,” and children are asked how much they are bullied. More commonly (e.g., Boivin & Hymel, 1997; Mynard & Joseph, 1997; Slee, 1994b), victimization experiences are dispersed across several items of a scale whose summed scores provide an index of victimization. As several different labels have been used to classify these examples of victimization, it is useful to consider the specific examples of experiences that have been used in single- and multiple-item measures of victimization. These examples are listed in Table 26.1 alongside the studies in which they were used, and the category in which they are grouped in the present review. Five categories (indirect, relational, physical, verbal, and generic victimization) are used to classify them, for the reasons described in the following. Recent European researchers have distinguished among physical, verbal, and indirect victimization (Björkqvist, 1994; Rivers & Smith, 1994). In North America Crick and colleagues (1999) have influentially distinguished physical and relational victimization. Relational victimization is similar to indirect victimization and both share some items in their operational definitions. But, as Crick et al. pointed out, indirect and relational aggression are conceptually distinct. Indirect aggression is defined as aggression which is enacted through a third party or so that the aggressor cannot be identified by the victim (Björkqvist, 1994). Relational aggression is defined as behavior which causes, or threatens to cause, damage to peer relationships, particularly to friendship and acceptance (Crick et al., 1999). In the present review and in Table 26.1, therefore, we have attempted to distinguish indirect and relational victimization experiences according to the preceding definitions.
Table 26.1 Form of Victimization Assessed in Cross-sectionAdjustment Research
Victimization Description of Category Target’s Experience Used in Operational Definition of Victimization Indirect Target sent nasty notes Lies told about target Mean things said to others about
Studies Using Description to Define Victimization
Boulton & Underwood (1992) Crick & Grotpeter (1996) Crick & Grotpeter (1996); Kochenderfer & Ladd (1996)
Annual progress in child psychiatry and child development 2000–2001
Relational
Physical
464
target Kupersmidt et al. (1997) Target talked about behind back Sharp (1996) Rumors spread about target Target not allowed Alsaker (1993);a Crick & Grotpeter (1996) to take part/be in Alsaker (1993);a Crick & Grotpeter (1996); group Kupersmidt et al. (1997) Target kept/left Boulton & Underwood (1992); Sharp (1996) out Crick & Grotpeter (1996) Target sent to Kupersmidt et al. (1997) Coventry/no one would talk to target Target told that another won’t like target unless target does what the other says Target rejected Target hit Alsaker (1993);a Austin & Joseph (1996); Boivin & Hymel (1997);a Boivin et al. (1995);a Boulton & Smith (1994); Boulton & Underwood (1992); Callaghan & Joseph (1995; self-reported victimization); Crick & Grotpeter (1996); Kochenderfer & Ladd (1996); Mynard & Joseph (1997); Neary & Joseph (1994; selfreported victimization); Rigby & Slee (1992); Sharp (1996); Slee (1994b, 1995b, 1995c); Slee & Rigby (1993b); Vernberg (1990)
Target pushed
Target kicked Target’s hair pulled Target locked inside a
Alsaker (1993);a Austin & Joseph (1996); Boivin & Hymel (1997);a Boivin et al. (1995);a Boulton & Underwood (1992); Callaghan & Joseph (1995; self-reported victimization); Crick & Grotpeter (1996); Mynard & Joseph (1997); Neary & Joseph (1994; selfreported victimization); Rigby & Slee (1992); Slee (1994b, 1995b, 1995c); Slee & Rigby (1993b); Vernberg (1990) Alsaker (1993);a Boulton & Underwood (1992); Crick & Grotpeter (1996); Sharp (1996) Alsaker (1993);a Crick & Grotpeter (1996)
Twenty Years’ research on peer victimization and psychosocial maladjustment
465
room Boulton & Underwood (1992) Others pick/start fights Byrne (1994); Olweus (1978)a with target Byrne (1994); Olweus (1978)a Others rough with target Crick & Grotpeter (1996); Vernberg (1990) Target shoved O’Moore & Hillery (1991) “Bullying…can be Sharp (1996) physical” Target’s belongings taken Verbal Target teased Austin & Joseph (1996); Boulton & Smith (1994); Byrne (1994); Callaghan & Joseph (1995; self-reported victimization); Mynard & Joseph (1997); Neary & Joseph (1994; selfreported victimization); Olweus (1978);a Vernberg (1990) Target laughed Austin & Joseph (1996); Byrne (1994); at/ridiculed Callaghan & Joseph (1995; self-reported victimization); Mynard & Joseph (1997); Neary & Joseph (1994; self-reported victimization); Olweus (1978)a Target called Austin & Joseph (1996); Callaghan & Joseph names/nasty names (1995; self-reported victimization); Mynard & Joseph (1997); Neary & Joseph (1994; selfreported victimization); Rigby & Slee (1992); Sharp (1996); Slee (1994b, 1995b, 1995c); Slee & Rigby (1993b) Nasty/unpleasant/mean Boulton & Underwood (1992); Kochenderfer & things said to target Ladd (1996) Target threatened Boulton & Underwood (1992); Sharp (1996) “Bullying…can be O’Moore & Hillery (1991) mental” GenericTarget picked on Austin & Joseph (1996); Callaghan & Joseph (1995; self-reported victimization); Kochenderfer & Ladd (1996); Kupersmidt et al. (1997); Mynard & Joseph (1997); Neary & Joseph (1994; selfreported victimization); Rigby & Slee (1992); Slee (1994b, 1995b, 1995c); Slee & Rigby (1993b); Vernberg (1990) Target bullied Austin & Joseph (1996); Callaghan & Joseph (1995; self-reported and peer-reported victimization); Kupersmidt et al. (1997); Mynard & Joseph (1997); Neary & Joseph (1994; self-reported and peerreported victimization); Rigby (1996); Slee (1994a)
Annual progress in child psychiatry and child development 2000–2001
466
Target harassed or Björkqvist et al. (1982);a Lagerspetz et al. tormented (1982) Others try to hurt Boivin & Hymel (1997);a Boivin et al. (1995) target’s feelings a a Mean/nasty things done Boivin & Hymel (1997); Boivin et al. (1995); Boulton & Smith (1994) to target Byrne (1994); Olweus (1978)a Target exposed to Slee (1994b, 1995b, 1995c); Slee & Rigby aggression (1993b) Target made fun of aThe reported description of victims’ experience was a translation into English of a term in another language. Alongside relational and indirect victimization, physical victimization is considered here as any form of victimization in which the victim’s physical integrity is attacked. Verbal victimization is considered victimization in which the victim’s status is attacked or threatened with words or vocalizations. There are some conceptual difficulties in labeling this form of victimization as “verbal,” in that words are also used to exclude victims (relational victimization) or to harm them through third parties (indirect victimization). But the types of experience grouped as “verbal” in Table 26.1 are frequently described as “verbal” in the literature (e.g., Björkqvist, 1994; Perry et al., 1988; Rivers & Smith, 1994); therefore, it is appropriate to describe them as such here. Finally, generic victimization denotes nonspecific descriptions of victimization experiences, which could include any of the other forms of victimization. In most of the cross-sectional studies reviewed here, correlational methods, or tests of difference between mean scores for victims, nonvictims, and other groups have been used to express the associations between one or more of these forms of peer victimization and psychosocial adjustment. The forms of psychosocial maladjustment included here are: depressed or dysphoric mood, loneliness, low social and global self-esteem and selfworth, and generalized and social anxiety. These are the principal forms of psychosocial maladjustment which have been investigated previously. It is not our intention to present an exhaustive review that would include other measures which might be described as psychosocial maladjustment. We will not include generalized measures of internalizing distress (such as general maladjustment and inadequacy, as assessed by Olweus, 1978), nonsocial, nongeneral aspects of children’s selfconcepts (such as academic, physical, or behavioral self-competence, as assessed by Austin & Joseph, 1996), measures related mainly to school adjustment (such as school liking, as assessed by Kochenderfer & Ladd, 1996), forms of adjustment which have rarely been investigated in cross-sectional studies (such as self-restraint, measured by Crick & Bigbee, 1998), or behaviorally oriented measures of internalizing problems (such as social withdrawal or submissiveness). Source of Informants and Shared Method Variance The informants who have been asked to make assessments of victimization in crosssectional studies have included not only the children in the cohort (giving self-reports),
Twenty Years’ research on peer victimization and psychosocial maladjustment
467
but also their teachers, parents, or peers. In some of the studies reviewed, victimization and adjustment have been assessed by asking questions of the same informants; for instance, both have been assessed by self-reports (e.g., Crick & Grotpeter, 1996). In other studies, separate informants have been used to measure separate variables; for example, Boivin and Hymel (1997) assessed victimization by peer report and loneliness by selfreport. Recently, several studies have compared the adjustment correlates of peerassessed and self-assessed victimization (Crick & Bigbee, 1998; Graham & Juvonen, 1998; Haselager, 1997). In these studies, self-reported victimization was more strongly associated than peer-reported victimization with self-re-ported maladjustment, and peer-reported victimization was more strongly associated than self-reported victimization with peerreported maladjustment. In other words, effect sizes were larger when informants were the same than when they were different. To explain such findings the investigators proposed a variety of explanations. The most parsimonious explanation and the most useful for present purposes, was suggested by Crick and Bigbee (1998) and Haselager (1997), and that is that the results are due to shared method variance. When the same method is used to assess outcome and predictor variables, any resulting correlation between outcome and predictor could be explained partly by the fact that measurement variance is shared between the two variables (e.g., Olweus, 1993b). Thus, a correlation between how unhappy children feel, and how victimized they say they are, may not primarily represent the association between victimization and unhappiness per se. Instead, it may represent the extent to which children who have negative feelings about one aspect of their life tend also to have negative feelings about another aspect. Or it may reflect the tendency of depressives to selectively recall negative events (cf. Hammen & Glass, 1975). In contrast, a correlation between children’s own feelings of unhappiness, and their degree of victimization as assessed by the reports of other informants, is not so open to such alternative interpretation. Other investigators in the field of peer relations have made similar points (e.g., Kupersmidt & Patterson, 1991), and recommended that outcome and predictor variables be assessed from different (or multiple) sources. When victimization is assessed by different informants from those assessing adjustment, in the studies reviewed here, it is assumed that shared method variance has been avoided as an alternative explanation of significant results. Summary of Research Aims Thus, in the present meta-analysis we aimed to collate the results of crosssectional studies that have made separate measurements of peer victimization (of generic, physical, relational, indirect, or verbal types) and psychosocial maladjustment (in the form of measures of depressive or anxious symptoms, loneliness, and negative global and social self-concepts). For the theoretical and empirical reasons outlined earlier, we predicted that victimization would be positively related to each of these forms of maladjustment. As well as summarizing effect sizes across studies, we aimed to note a number of their attributes which seemed important for an evaluation. These included participants’ sex, age groups, and nationalities; the subtypes of victimization measured; the source of informants (self-, peer, teacher, or parent reports); the presence of shared method
Annual progress in child psychiatry and child development 2000–2001
468
variance in effect sizes; and the nature of adjustment measurements used. As shared method variance potentially has a major impact on the magnitude of effect sizes, we separated effect size summaries according to the degree of shared method variance present in measurements.
METHOD Literature Search We attempted to locate all cross-sectional studies of peer victimization and psychosocial maladjustment published between 1978, when the influential work of Olweus appeared, and the end of June 1997. Although it would have been interesting to add more recent studies, a great number of cross-sectional investigations were already published by that date. As they are continually being added to, it was necessary to place a limit on the time period to be sampled. Published studies were located using a variety of recursive methods. We searched electronic databases, PsycLit, BIDS ISI Social Science Citation Index, and OCLC Firstsearch, using keywords “bully*” and “vict*” and also the names of known bullying and peer victimization researchers. We consulted Skinner’s (1996) bibliography of bullying literature. We sought relevant publications cited in reviews of the bullying literature and elsewhere in the peer relations literature. We followed up citations, in articles and book chapters retrieved by these methods, that referred to other work on victims’ adjustment. We conducted hand searches of journals in which papers on victims’ adjustment had previously been published, and we wrote to authors who had published in the field, asking for any relevant publications of which we were not aware. Manuscripts retrieved in this way which remained unpublished by June 1997 were not included in the meta-analysis. Calculation of Effect Sizes. This meta-analysis followed the procedures outlined by Rosenthal (1984) and Strube (1985) for calculating effect sizes, based on Pearson’s correlation coefficient (r), and standard normal deviates (Z-scores), for each study, and one-tailed probability values (as recommended by Rosenthal) for mean effect sizes. Pearson’s r was chosen as the measure of effect size (Rosenthal, 1984) for summarizing studies in this meta-analysis. These rs, based on correlations between a continuum of victimization and psychosocial maladjustment, amounted to a comparison of victims with nonvictims. Direct estimates of r were avaiable from some studies as Pearson’s rs (Alsaker, 1993; Austin & Joseph, 1996; Boivin & Hymel, 1997; Boivin et al., 1995; Callaghan & Joseph, 1995, for maladjustment of self-assessed victims; Kochenderfer & Ladd, 1996; Mynard & Joseph, 1997; Neary & Joseph, 1994, for maladjustment of self-assessed victims; Rigby & Slee, 1992; Slee, 1994a, 1994b, 1995c; Vernberg, 1990), multiple Rs (Crick & Grotpeter, 1996), or εs (Olweus, 1978). When rs were not available (Björkqvist et al., 1982; Boulton & Smith, 1994; Byrne, 1994; Callaghan & Joseph, 1995, for maladjustment of peer-assessed victims; Neary & Joseph, 1994, for maladjustment of peer-assessed victims; O’Moore & Hillery, 1991; Sharp, 1996; Slee & Rigby, 1993b), we
Twenty Years’ research on peer victimization and psychosocial maladjustment
469
computed rs from the statistics reported by the authors, according to formulae provided by Rosenthal (1984). In some studies (Björkqvist et al., 1982; Boulton & Smith, 1994; Lagerspetz, Björkqvist, Berts, & King, 1982; Olweus, 1978) victims were compared with tightly defined groups such as bullies, welladjusted children, or not-involved children. In these studies, effect sizes (ωs, equivalent to rs—see Howell, 1992) were calculated from the omnibus Fs (if available) from analyses in which data from all groups of participants had been included. Omnibus Fs were not avaialble in the studies published by Lagerspetz et al. (1982) and Slee and Rigby (1993b), but within-group means and standard deviations were. From these, rs were computed via omnibus Fs (for Lagerspetz et al., 1982), and Hedge’s g (Rosenthal, 1984; for Slee & Rigby, 1993b). Mean effect sizes were calculated separately for studies in which the shared method variance was, and was not, present in the effect. The first of the present authors calculated Z-statistics, and mean effect sizes and significance levels, using formulae provided by Rosenthal (1984) and Strube (1985), and tables for calculating Fisher’s transformation of r to transformed r (necessary for the computation of unbiased mean effect sizes) from Howell (1992). When effect sizes were reported separately for more than one independent group of participants (e.g., males and females in Slee, 1994a,b, 1995b,c), the means of the rs and Zs reported or computed were displayed in the meta-analysis. When more than one test of the same hypothesis was carried out, within a single study, on a single group of participants (Boivin et al., 1995; Kochenderfer & Ladd, 1996; Slee, 1994b; Vernberg, 1990), mean rs were calculated using Rosenthal’s formula, and mean Zs using Strube’s adjusted formula for nonindependence of hypotheses. The same procedure was followed to combine effect sizes across subtypes of victimization, from studies in which effects were reported for more than one subtype (Alsaker, 1993; Crick & Grotpeter, 1996). In these instances, correlations between repeated measures were used in adjusting for dependence, as recommended by Strube. There was one instance in which different analyses of the same data set had been published separately (Boivin & Hymel, 1997; Boivin et al., 1995). When effect sizes could be estimated from both publications (as for loneliness) they were taken from the study with the larger data set (Boivin & Hymel, 1997). When more than one test had been carried out, but full details (i.e., statistical values and, if necessary, the sample size) had not been reported (Björkqvist et al., 1982; Callaghan & Joseph, 1995; Olweus, 1978), the first author computed the smallest possible effect size from the results available. This was not possible in Callaghan and Joseph’s study, as they published only the maximum effect sizes in their report, and so these were used instead.
RESULTS Overview of Study Attributes Variety of Participants. The victimization-adjustment association has been investigated among an impressive variety of populations. Both boys and girls have been considered in most studies, although only girls were included in Neary and Joseph’s (1994) study, and only boys in two studies (Olweus, 1978; Slee & Rigby, 1993b). The age range of children
Annual progress in child psychiatry and child development 2000–2001
470
studied is also broad: it has included infant and preschool children (Alsaker, 1993; Kochenderfer & Ladd, 1996) and adolescents (Björkqvist et al., 1982; Lagerspetz et al., 1982; Olweus, 1978; Rigby & Slee, 1992; Slee, 1994a, 1995b; Vernberg, 1990). But most of the studies’ participants have been in their middle childhood (aged between 8 and 13 years: Austin & Joseph, 1996; Boivin & Hymel, 1997; Boivin et al., 1995; Boulton & Smith, 1994; Callaghan & Joseph, 1995; Crick & Grotpeter, 1996; Mynard & Joseph, 1997; Neary & Joseph, 1994; O’Moore & Hillery, 1991; Sharp, 1996; Slee, 1994b, 1995c, Slee & Rigby, 1993b). Participants have been drawn from a variety of countries. These include Australia (Rigby, 1996; Rigby & Slee, 1992; Slee, 1994a, 1994b, 1995a, 1995b, 1995c; Slee & Rigby, 1993a, 1993b); French Canada (Boivin & Hymel, 1997; Boivin et al., 1995); Northern Ireland (Callaghan & Joseph, 1995); the Irish Republic (Byrne, 1994; Neary & Joseph, 1994; O’Moore & Hillery, 1991); Finaland (Björkqvist et al., 1982; Lagerspetz et al., 1982); Norway (Alsaker, 1993); Sweden (Olweus, 1978); the United States (Crick & Grotpeter, 1996; Kochenderfer & Ladd, 1996; Vernberg, 1990); and mainland Britain (Austin & Joseph, 1996; Boulton & Smith, 1994; Boulton & Underwood, 1992; MacLeod & Morris, 1996; Mynard & Joseph, 1997; Sharp, 1996; Williams, Chambers, Logan, & Robinson, 1996). With the exception of two studies of a single Frenchspeaking group of children (Boivin & Hymel, 1997; Boivin et al., 1995), all the participants have been English-speaking or Scandinavian. Subtypes of Victimization. In most of the studies reviewed, subtypes of victimization were not assessed separately. Rather, victimization has been measured as a composite of two or more subtypes (Austin & Joseph, 1996; Björkqvist et al., 1982; Boivin & Hymel, 1997; Boivin et al., 1995; Boulton & Smith, 1994; Byrne, 1994; Callaghan & Joseph, 1995; Kochenderfer & Ladd, 1996; Lagerspetz et al., 1982; Mynard & Joseph, 1997; Neary & Joseph, 1994; Olweus, 1978; O’Moore & Hillery, 1991; Rigby & Slee, 1992; Slee, 1994a, 1994b, 1995b, 1995c; Slee & Rigby, 1993b; Vernberg, 1990). Physical and verbal victimization (or at least forms of victimization approximating their definitions) have generally been included in these victimization composites, or when subtypes were distinguished. Assessments of victimization have included relational or indirect victim-ization in only five of the published studies reviewed (Alsaker, 1993: relational victimization only; Crick & Grotpeter, 1996; Kochenderfer & Ladd, 1996: indirect victimization only; Sharp, 1996; and Boulton & Underwood, 1992, in which data for the calculation of effect sizes were not published). Additionally, in some studies (Björkqvist et al., 1982; Lagerspetz et al., 1982; Slee, 1994a) only a generic form of victimization has been assessed, without further specification of the types of victimization that are meant. Source of Informants. Victimization has most commonly been assessed by self-report or peer report (e.g., Austin & Joseph, 1996; Björkqvist et al., 1982; Boivin & Hymel, 1997; Boivin et al., 1995; Boulton & Smith, 1994; Callaghan & Joseph, 1995; Crick & Grotpeter, 1996; Kochenderfer & Ladd, 1996; Lagerspetz et al., 1982; Mynard & Joseph, 1997; Neary & Joseph, 1994; O’Moore & Hillery, 1991; Rigby & Slee, 1992; Sharp, 1996; Slee, 1994a, 1994b, 1995b, 1995c; Slee & Rigby, 1993b; Vernberg, 1990). Psychosocial maladjustment has usually been measured by self-report (e.g., Austin & Joseph, 1996; Björkqvist et al., 1982; Boivin & Hymel, 1997; Boivin et al., 1995;
Twenty Years’ research on peer victimization and psychosocial maladjustment
471
Boulton & Smith, 1994; Byrne, 1994; Callaghan & Joseph, 1995; Crick & Grotpeter, 1996; Kochenderfer & Ladd, 1996; Lagerspetz et al., 1982; Mynard & Joseph, 1997; Neary & Joseph, 1994; O’Moore & Hillery, 1991; Rigby & Slee, 1992; Sharp, 1996; Slee, 1994a, 1994b, 1995b, 1995c; Slee & Rigby, 1993b; Vernberg, 1990). Effects of Victimization on Contemporaneous Maladjustment Overview. Tables 26.2 to 26.6 summarize the results of published studies of the relationship between victimization and each form of psychosocial maladjustment. The number and age range of participants contributing to the effect size are displayed in the second and third columns of each table. Unless otherwise indicated, children of both sexes participated in the studies. In the fourth column is an indications of the type of informants’ reports used to assess victimization (self-, peer, teacher, or parent reports). In the fifth column is a description of the subtypes of victimization assessed, as far as can be determined from the authors’ descriptions, and according to the classification laid out in Table 26.1 (generic, physical, verbal, relational, or indirect). These are noted in abbreviated form, as indicated in the footnotes to each table. The effects in each table are separated according to whether there was shared method variance due to shared informants present in the effect. For example, if both victimization and maladjustment were assessed by self-report, shared method variance was not avoided as an explanation of effect size. If victimization was assessed by peer report, and maladjustment by self-report, shared method variance was avoided as an explanation of effect size. In the final columns of Tables 26.2 to 26.6 the effect size r and the corre-spending Zscore are reported, within cash category of effect (with and without shared method variance) studies are listed in order of increasing effect size. Mean effect sizes and their overall significance levels are shown separately for studies with and without shared method variance. In the display of effect sizes, the valence of negative rs and Zs is reversed when these indicate a negative relationship between victimization and maladjustment. Thus, the effect sizes presented in Tables 26.2 to 26.6, which were all positive, represent the magnitude of positive associations between victimization and maladjustment found in each study. One-tailed significance levels, as recommended by Rosenthal (1984), are given for the overall Z-score, but not for individual Z-scores, because it is the overall significance level which is of interest.
TABLE 26.2 Published Studies of the Contemporaneous Association between Victimization and Depression
Age Range Victimization Study N (Years) Informants With shared method variance Vernberg (1990) 73 12–14 Self Slee (1995b) 220 12–17 Self Slee (1994a) 363 12–15 Self
Victimization Subtypesa GPV GPV G
r
Z
.23 1.80 .26 3.86 .31 5.72
Annual progress in child psychiatry and child development 2000–2001
Björkqvist et al. 67 14–16 (1982) Austin & Joseph 425 8–11 (1996) Crick & 438 8–12 Grotpeter Slee (1995c) 290 M=10.9 (1996) Callaghan & 120 10–12 Joseph (1995) 10–12 Neary & Joseph 60 c (1994) Mean effect size Avoiding shared method variance Boivin et al. 567 9–12 (1995) Callaghan & 63 10–12 Joseph (1995) 10–12 Neary & Joseph 60 (1994)c Mean effect size
472
Peers
G
.38 2.33
Self
GPV
.39 7.93
Self
PRI
.42 8.79
Self
GPV
.51 8.60
Self
GPVb
.53 4.21
Self
GPVb
.81 6.27 .45d
Peers
GPV
.24 5.41
Peers
Gb
.26 2.85
Peers
Gb
.36 2.79
.29d aG=generic victimization; I=indirect victimization; P=physical victimization; R=relational victimization; V=verbal victimization. bSeparate effect sizes were published for peer-reported and self-reported victimization. cGirls only. dp<.0001.
Depression. It is clear from the studies presented in Table 26.2 that victimization is positively associated with depression. Mean effect sizes were significantly greater than zero (p<.0001) whether or not there was shared method variance in the effect. Victims of peer aggression tended to be more depressed than nonvictims. Effect sizes were smaller when shared method variance was avoided than when it was not. When victimization was assessed by peers, and depression by self-report, the mean effect size was r=.29, representing 8.4% shared variance. The range was .24 to .36. When both victimization and depression were assessed by self-report, the mean effect size was r=.45 (20.3% shared variance), with a range of .23 to .81. Nevertheless, victimization and depression were clearly associated independently of shared method variance. Similar findings have been made in research which is not summarized in Table 26.2, and which used slightly different dependent measures. MacLeod and Morris (1996) reported that over 4% of a sample of children who called a bullying telephone helpline head expressed suicidal thoughts. These authors did not make a comparison to the
Twenty Years’ research on peer victimization and psychosocial maladjustment
473
prevalence of suicidal thoughts in the normal population, or among callers to helplines that are not specifically targeted at victims of peer aggression, however. Rigby (1996) found that among secondary school children victims were twice as likely as nonvictims to report suicidal thoughts and other symptoms of depression.1 Other researchers have also found that victims rate themselves as generally less happy (Rigby & Slee, 1992; Williams et al., 1996), or less happy in a school context (Boulton & Underwood, 1992; Slee, 1995a, 1995c; Slee & Rigby, 1993a), than nonvictims. Data concerning unhappiness and suicidal ideation are not included in Table 26.2, as there was an abundance of studies that used well-validated measures of depression, such as the Children’s Depression Inventory (Kovacs, 1992—used by Boivin et al., 1995; and Crick & Grotpter, 1996) and the Depression SelfRating Scale (Birleson, 1981—used by Austin & Joseph, 1996; Callaghan & Joseph, 1995; Neary & Joseph, 1994; and Slee, 1995c). The major limitation of depression-victimization research is in its measurement of victimization. Only one study measured relational or indirect victimization (Crick & Grotpeter, 1996), and this study did not include verbal victimization. Only one study (Boivin et al., 1995) used more than one item to assess peer-reported victimization. Loneliness. Fewer studies have been published in which loneliness was the dependent variable. In these studies the mean effect size from published studies was smaller for loneliness than for depression. Nevertheless, it is clear from the studies presented in Table 26.3 that loneliness is positively associated with victimization. Mean effect sizes were significantly greater than zero (p<.0001), whether or not shared method variance was avoided. The tendency was, again, for effects to be smaller when shared method variance was avoided (mean r= .25, range=.15 to .34) than when it was not (mean r=.32, range=.14 to .49). On average, loneliness and victimization shared 6.3% of variance when there was no shared method variance, and 10.2% of variance when there was shared method variance. Thus, victims were more lonely than nonvictims, irrespective of shared method variance. Similar conclusions have been drawn from unpublished data cited in published reports (Boulton & Underwood, 1992; Slee & Rigby, 1994), and can be drawn from an analysis of Boivin and Hymel’s (1997) data with a smaller sample (Boivin et al., 1995). Most of the studies (Boivin & Hymel, 1997; Crick & Grotpeter, 1996; Slee & Rigby, 1994) employed a well-validated measure of loneliness—the Loneliness and Social Dissatisfaction Scale (Asher & Wheeler, 1985), or its equivalent for younger children (used by Kochenderfer & Ladd, 1996). It is possible that other measures of loneliness are not so strongly correlated with victimization, given the smaller effect sizes calculated from Alsaker’s (1993) data. Diverse forms of victimization are fairly well represented in effect sizes, although none can be calculated from victimization measures which comprised all four of physical, verbal, relational, and indirect victimization. 1
Rigby (1996) did not report total depression scores that would allow computation of effect sizes.
Annual progress in child psychiatry and child development 2000–2001
474
TABLE 26.3 Published Studies of the Contemporaneous Association between Victimization and Loneliness
Age Range Victimization Informants Study N (Years) With shared method variance Alsaker (1993) 120 6–7 Self, teachers, peers, parents Kochenderfer & 200 5–6 Self Ladd (1996) Crick & Grotpeter 438 8–12 Self (1996) Mean effect size Avoiding shared method variance Alsaker (1993) 120 6–7 Self, teachers, peers, parents Boivin & Hymel 798 8–10 Peers (1997) Mean effect size
Victimization Subtypesa
r
Z
PVR
.14 1.50
GPVI
.31 4.76
PRI
.49 10.25 .32b
PVR
.15 1.66
GPV
.34 9.57
.25b aG=generic victimization; I=indirect victimization; P=physical victimization; R=relational victimization; V=verbal victimization. bp<.0001.
Anxiety and Social Anxiety. In the literature it is fairly common to find victimization positively correlated with some measure of social anxiety (e.g., Alsaker, 1993; Boulton & Smith, 1994; Crick & Grotpeter, 1996; Slee, 1994b), or with constructs that are similar but not equivalent to anxiety, such as neuroticism (Byrne, 1994; Mynard & Joseph, 1997; Slee & Rigby, 1994) or anxious self-concept (O’Moore & Hillery, 1991). Studies of the relationship between victimization and generalized anxiety are less common. In two separate studies, Slee (1994a, 1995b) found that victimization was positively correlated with anxiety measured as a subscale of a published health symptom checklist. Using the same items from this checklist (but not combining them in a single index of anxiety), Rigby (1996) also found more symptoms of anxiety among victims than nonvictims. Olweus (1978) and Lagerspetz et al. (1982) found that victimization was positively correlated with unvalidated measures of anxiety. In unpublished studies, Pierce (1990) and Kupersmidt, Voegler, Sigda, and Sedikides (1997) have also shown that anxiety was positively related to victimization.
Twenty Years’ research on peer victimization and psychosocial maladjustment
475
TABLE 26.4 Published Studies of the Contemporaneous Association Between Victimization and Anxiety
Age Range Victimization Study N (Years) Informants Generalized Anxiety With shared method variance Slee (1995b) 220 12–17 Self Slee (1994a) 363 12–15 Self Mean effect size Avoiding shared method variance Lagerspetz et 239 12–16 Peers al. (1982) 64 13 Teachers Olweus b (1978) Mean effect size Social Anxiety With shared method variance Alsaker 120 6–7 Self, teachers, (1993) peers, parents Boulton & 57 8–10 Peers Smith (1994) Slee (1994b) 114 9–13 Self Crick & 438 Grotpeter (1996) Crick & 438 Grotpeter (1996) Slee (1994b) 114
Victimization Subtypesa
r
Z
.20 2.97 .29 5.53 .2d
GPV G
Anxiety Anxiety
G
Q-inventory .18 2.78 anxiety Q-sort .24 1.67 anxiety
PV
.21c
PVR
Fear of peers .13 1.03
GPV
Shyness
.17 1.28
GPV
Social avoidance Social anxiety
.25 2.67
8–12
Self
PRI
8–12
Self
PRI
9–13
Self
Mean effect size Avoiding shared method variance Alsaker 120 6–7 Self, teachers (1993) Overall meanShared effect method
Dependent Measure
GPV
.26 5.44
Social avoidance
.30 6.28
Fear of negative evaluation
.40 4.27 .25d
PVR
Fear of peers, .14c2.50 peers, parents .25c
Annual progress in child psychiatry and child development 2000–2001
476
variance .19c No shared method variance aG=generic victimization; P=physical victimization; V=verbal victimization; R=relational victimization; I=indirect victimization. bBoys only. cp<.01; dp<.0001. sizes for anxiety
The studies are summarized separately for social and generalized anxiety in Table 26.4, with an indication of the measures of anxiety used. Table 26.4 shows that the mean effect sizes for social and generalized anxiety were barely different from one another. Victims were more generally and socially anxious than nonvictims, independently of shared method variance. The effect sizes were smaller than for other forms of psychosocial maladjustment, although they were again clearly greater than zero (p<.01). Across studies in which there was shared method variance, victimization shared 6.3% of its variance with both social and general anxiety. When there was no shared method variance, victimization shared 4.3% of its variance with general anxiety and 2.0% of its variance with social anxiety. There are relatively few of these studies in which validated measures of generalized anxiety have been employed, and in which the breadth of victimization experience has been assessed. Only one, unpublished, study (Kupersmidt et al., 1997) measured physical, verbal relational, and indirect victimization. General or Global Self-Esteem. A large number of cross-sectional studies have investigated the relation between peer victimization and children’s global or general selfconcepts. Across these studies, as shown in Table 5, victimization was correlated with low self-esteem, independently of shared method variance (mean r=.21, p<.0001, representing 4.4% shared variance), and more so in studies in which shared method variance was not avoided (mean r=.39, p<.0001, 15.2% shared variance). It is possible that shared method variance may bias the mean effect sizes across self-concept studies more than in the studies involving depression, loneliness, and anxiety. This bias is due to the results of four sudies by Joseph and his colleagues (Austin & Joseph, 1996; Callaghan & Joseph, 1995; Mynard & Joseph, 1997; Neary & Joseph, 1994). Neary and Joseph (1994) developed a self-report peer victimization scale, which they immersed in Harter’s (1985) Self-Perception Profile for Children such that children were asked to respond to its items in the same way as they did to other items in Harter’s scale. The same peer victimization scale was used in the other work published by Joseph and colleagues. In other words, the methods of assessing victimization and self-concept used by Joseph have even more in common than they do in other studies where both variables were selfassessed (such as Alsaker, 1993; O’Moore & Hillery, 1991; Rigby & Slee, 1992; Sharp, 1996; Slee & Rigby, 1993b). Sure enough, in Joseph’s work, there are high correlations among the subscales of Harter’s measure, and also between these and Joseph’s peer victimization scale. The mean r for global self-concept across his four studies
Twenty Years’ research on peer victimization and psychosocial maladjustment
477
summarized in Table 5 was .48. In contrast, the mean r for global self-concept across the other published studies in which there was shared method variance (Alsaker, 1993; O’Moore & Hillery, 1991; Rigby & Slee, 1992; Sharp, 1996; Slee & Rigby, 1993b) was .31. All mean rs remained highly significant (p<.0001). Leaving aside these considerations, one of the strengths of the cross-sectional studies of global self-concepts is that they have employed a variety of different, widely used, and empirically validated self-esteem inventories, and consistently found that victims have lower self-esteem than other children. One of the limitations is that relational and indirect victimization, according to authors’ reports, have only been assessed by Sharp (1996; relational and indirect victimization) and Alsaker (1993; relational victimization). Another is that, when shared method variance has been avoided, only one item has been used to measure victimization (except by Byrne, 1994). Social Self-Concept. Measures of children’s social self-concept index the extent to which they see themselves as being socially competent, well-accepted by their peers, or having good social relationships. It has generally been found that victims tend to have negative views of themselves in the social domain, as shown in Table 26.5. Again, this pattern has been shown with a variety of widely used and validated instruments. The mean effect size for published studies with shared method variance was .35 (12.3% variance shared between victimization and social self-esteem), and .23 for those without (5.3% variance shared between the variables); both were highly significant (p<.0001). The magnitude of effect sizes in studies with shared method variance was again affected by the measure of victimization used. Across Joseph’s studies (Austin & Joseph, 1996; Callaghan & Joseph, 1995; Mynard & Joseph, 1997; Neary & Joseph, 1994) the mean effect size for self-reported victimization and selfreported social self-worth was .46. Across the remaining studies (O’Moore & Hillery, 1991; Slee & Rigby, 1993b; Vernberg, 1990) that relied exclusively on self-reports the mean effect was .20. A major limitation of this group of studies is that none of them employed a composite measure of victimization that included either relational or indirect victimization.
TABLE 26.5 Published Studies of the Contemporaneous Association between Victimization and Global/General Self-Esteem
Age Range Study N (Years) With shared method variance O’Moore & 783 7–13 Hillery (1991) Rigby & Slee 810 12–18 (1992) Alsaker (1993) 120 6–7 Austin & Joseph (1996) Sharp (1996)
Victimization Informants
Victimization Subtypesa
r
Z
Self
G
.12 3.36
Self
GPV
.22 6.26
PVR
.24 5.13
GPV
.38 7.70
PVRI
.41 7.96
425
8–11
Self, teachers, peers, parents Self
377
11–12
Self
Annual progress in child psychiatry and child development 2000–2001
Mynard & 179 8–13 Joseph (1997) 87 7–13 Slee & Rigby b (1993b) 10–12 Neary & Joseph 60 c (1994) Callaghan & 120 10–12 Joseph (1995) Mean effect size Avoiding shared method variance Alsaker (1993) 120 6–7 Lagerspetz et al. (1982) Boulton & Smith (1994) Byrne (1994)
478
Self
GPV
.45 6.02
Self
GPV
.52 4.85
Self
GPV
.53 4.11
Self
GPV
.55 4.26 .39d
239
12–16
Self, teachers, peers, parents Peers
76
8–10
Peers
PVR
.03 1.75
G
.17 2.63
GPV
.17 1.51
177
Primary & Teachers, peers PV .23 3.00 secondary school age 13 Teachers PV .26 2.08 Olweus (1978)b 64 10–12 Peers G .22 1.70 Neary & Joseph 60 (1994)c Callaghan & 63 10–12 Peers G .38 2.94 Joseph (1995) Mean effect .21d size aG=generic victimization; P=physical victimization; V=verbal victimization; R=relational victimization; I=indirect victimization. bBoys only. cGirls only. dp<.0001. TABLE 26.6 Published Studies of the Contemporaneous Association between Victimization and Social Self-esteem
Age Range Victimization Study N (Years) Informants With shared method variance O’Moore & 783 7–13 Self Hillery (1991)
Victimization Subtypesa G
r
Z
.14 3.92
Twenty Years’ research on peer victimization and psychosocial maladjustment
Vernberg (1990) 73 12–14 87 7–13 Slee & Rigby b (1993b) Austin & Joseph 425 8–11 (1996) 10–12 Neary & Joseph 60 c (1994) Callaghan & 120 10–12 Joseph (1995) Mynard & Joseph 179 8–13 (1997) Mean effect size Avoiding shared method variance Boulton & Smith 76 8–10 (1994) Callaghan & 63 10–12 Joseph (1995) Boivin & Hymel 793 8–10 (1997) 10–12 Neary & Joseph 60 c (1994) Mean effect size
479
Self Self
GPV GPV
.19 1.47 .26 2.43
Self
GPV
.39 9.54
Self
GPV
.43 3.33
Self
GPV
.49 5.37
Self
GPV
.52 6.96 .35d
Peers
GPV
.07 0.59
Peers
G
.23 1.83
Peers
GPV
.26 7.32
Peers
G
.34 2.63
.23d aG=generic victimization; P=physical victimization; V=verbal victimization. bBoys only. cGirls only. dp<.0001.
SUMMARY AND DISCUSSION This paper presented the first meta-analysis of the victimization-adjustment literature. The pattern of results across cross-sectional studies strongly suggested that victims of peer aggression experience more negative affect, and negative thoughts about themselves, than other children. The strength of previous research is in the number of studies that have been carried out, using a variety of methods and with participants drawn from diverse populations. For ease of comparison of the impact of different adjustment variables and differ-ent informants, mean effect sizes from Tables 26.2 to 26.6 are summarized in Table 26.7. Table 26.7 demonstrates clearly that effect sizes were uniformly larger when there was shared method variance than when there was not. In other words, victimization was more strongly correlated with psychosoical maladjustment when both variables were assessed by asking the same informants (who were usually the participants themselves), than when different informants were used to assess each
Annual progress in child psychiatry and child development 2000–2001
480
variable. Victimization and Different Forms of Maladjustment The relative sizes of effects for different forms of maladjustment are worthy of comment. Several authors have asserted that victims are typically fearful and anxious (e.g. Besag, 1989; Olweus, 1993a; Rigby, 1996; Ross, 1996; Tattum & Tattum, 1992). Similarly, it is widely asserted in the bullying literature that victims have low self-esteem (e.g., by Besag, 1989; Farrington, 1993; Kochenderfer & Ladd, 1996; Olweus, 1993a; Perry, Perry, & Kennedy, 1992; Randall, 1996; and Ross, 1996). In contrast, the largest effect sizes in this meta-analysis were for depression, and the smallest for anxiety. Effect sizes for loneliness, and social and global or general self-esteem, were midway between these. Additionally, effects for social and global self-esteem were substantially lowered after excluding a group of studies (Austin & Joseph, 1996; Callaghan & Joseph, 1995; Mynard & Joseph, 1997; Neary & Joseph, 1994) that used similarly structured victimization and self-concept measures. All together these results suggest that, although victims are indeed generally and socially anxious and have low global and social self-esteem, they are even more strongly characterized by feelings of loneliness and dysphoria.
TABLE 26.7 Summary of Published Studies of the Associations between Victimization and Psychosocial Maladjustment
Mean Effect Sizes (rs) Dependent Variable For Studies Avoiding Shared For Studies with Shared Method Variance Method Variance Depression .29 .45 Loneliness .25 .32 Global self-esteem .21 .39 Social self-concept .23 .35 Social anxiety .25 .14a .21* .25 Generalized anxiety Anxiety overall (social/ .19 .25 generalized) ap<.01. All other rs significant, p<.0001. This is an important observation, and one which deserves further investigation. These dysphoric feelings may find an explanation in social rank theory (Gilbert, 1992), which proposes that social experiences like victimization are implicated particularly in the development of depression. The associations of victimization with anxiety and a poor self-concept were first observed in pioneering research (Olweus, 1978), in which depression and loneliness were not assessed. One might hypothesize that these associations, being less strong across studies, are due primarily to the comorbidity of anxiety and low-self-esteem with depression. Unfortunately, the studies reviewed here do
Twenty Years’ research on peer victimization and psychosocial maladjustment
481
not allow a test of this depression hypothesis. To test the hypothesis, it is necesary to use more than one adjustment variable as a predictor of concurrent victimization, whereas controlling for the correlations between different forms of adjustment. In a recent study (Hawker & Boulton, 1998) we compared self-reported depressed mood, loneliness, anxiety, and social and global self-esteem as predictors of contemporaneous peer victimization among children aged 8 to 12 years. When intercorrelations between different forms of maladjustment were controlled, all of them were related to selfreported victimization, but depression and loneliness emerged as the strongest predictors of peer-reported victimization. These results offer some tentative support to the depression hypothesis. Replications using similar designs would help show whether victimization is primarily related to children’s negative thoughts about themselves and their lives, to their socially related concerns, to their anxiety-related cognitions, or to some combination of these. Another type of comparison available in Table 26.7 is between forms of maladjustment that might be described as primarily social (loneliness, social self-concept, and social anxiety) and those that might be described as primarily psychological (depression, global self-concept, and generalized anxiety). Given that peer victimization is a social experience, one might expect it to be more strongly related to social than to psychological forms maladjustment. In fact there is no evidence for this. If anything, it seems the other way around. Across the studies, victimization was no more strongly related to social selfworth than to global self-worth. Victimization was marginally less strongly related to social anxiety than to generalized anxiety, and less strongly related to loneliness than to depression. It would be unwise, given the variation in adjustment measures, to make too much of the greater effects for psychological maladjustment. But these results make it clear, at least, that victims’ emotional suffering is not confined to the social domain. Again, it is difficult to say whether victimization is primarily related to social or to psychological forms of maladjustment, without conducting a study which controls for the relation between different forms of maladjustment.
GENERAL IMPLICATIONS FOR FUTURE RESEARCH A number of more specific limitations of the studies reviewed here could be addressed in future research. First, few previous studies have included relational or indirect victimization, either in their own right (e.g., Alsaker, 1993; Crick & Grotpeter, 1996; Kochenderfer & Ladd, 1996), or as part of an index of composite victimization (e.g., Boulton & Underwood, 1992; Sharp, 1996). Even when one of these forms of victimization was assessed, either the other form or verbal victimization usually was not. It is possible that effect sizes have been exaggerated or underestimated as a consequence. Second, although in several studies multiple items have been used in the self-assessment of victimization, there is a shortage of studies that have used more than one item in peer assessments of victimization. Third, researchers have generally reported significance levels but not effect sizes. This practice may encourage the impression that a single study can show, on the basis of an arbitrary significance level, whether or not victimization is related to a particular form of maladjustment. We have attempted to show in this review
Annual progress in child psychiatry and child development 2000–2001
482
that it is more effective to combine effect sizes across studies than to base conclusions on the results of significance tests in a single study. Fourth, there is a need for studies of children from cultural backgrounds that are not Scandinavian or English-speaking. Fifth, it would be valuable to investigate moderating effects on victims’ adjustment of variables such as disability, age, gender, and sexual orientation. A recent example, using a slightly different methodology, is Rivers’ (in press) retrospective study of former victims reports of adjustment as a function of their sexuality. In many ways, however, the strengths of cross-sectional studies outweigh their weaknesses. Together they demonstrate that victims of peer aggression suffer a variety of feelings of psychosocial distress. They feel more anxious, socially anxious, depressed, lonely, and worse about themselves than do nonvictims. The evidence suggests that these feelings occur among victims of both sexes, all age groups, and all subtypes of aggression. Across studies in which different informants’ reports were used to measure victimization and adjustment, the aggregated effects show that victims’ reports of distress cannot be explained away as an artefact of shared method variance. That is, children who are seen as victims by their peers tend to report greater distress than children who are not seen as victims by their peers. Conclusions such as these have been drawn before from single empirical studies or in qualitative review papers. Here they are clearly demonstrated in a pattern of aggregated quantitative effects. They are not pleasant conclusions; they reveal a pattern of distress that can no longer be ignored. We suggest that the strongest implication of these conclusions is that there is little need now for further cross-sectional studies of peer victimization and psychosocial maladjustment. It is clear enough already that victims are dis-tressed. There is some value in designing studies which will address some of the limitations of past research, by accounting for issues such as comorbidity, the measurement of victimization, and crosscultural variability. Otherwise it is time for victimization research to move on. Early studies of bullying and victimization, ably reviewed by Farrington (1993), answered questions about prevalence. More recent studies, including those reviewed here, focused on the correlates of victimization. We suggest that it is time for researchers to look at questions that arise out of the conclusions of the current meta-analysis. Among these questions are matters of risk, causation, the differential relation of maladjustment to subtypes of victimization, the role of victimization in clinical presentations, and interventions to reduce victims’ distress. Some studies have been published which begin to address these questions (e.g., Crick & Grotpeter, 1996; Cunningham et al., 1998; Kochenderfer & Ladd, 1996; Olweus, 1993a, 1993b; Salmon, James, Cassidy, & Javaloyes, 1999; Smith & Sharp, 1994). However, they are far fewer in number than the cross-sectional studies incorporated in this meta-analysis. If more researchers devote themselves to addressing these more complex questions, then practitioners may being to make a serious impact on the distress that children feel when they are bullied.
ACKNOWLEDGMENTS This work was completed with financial assistance from the University of Keele. Portions of the material were presented at the biennial meeting of the International Society for the
Twenty Years’ research on peer victimization and psychosocial maladjustment
483
Study of Behavioral Development, Berne, July 1998, and to the British Psychological Society, London, December 1998. We are grateful to Precilla Y.L.Choi, Susan A.O’Neill, and Peter K.Smith for their comments on earlier drafts of this manuscript.
REFERENCES1 *Alsaker, F.D. (1993). Isolement et maltraitance par les pairs dans les jardins d’enfants: comment mesurer ces phénomènes erquelles sont lours conséquences? [Isolation and peer abuse in day-care centres: How can these phenomena be measured and what are their consequences?] Enfance, 47, 241–260. Ambert, A.-M. (1995). Toward a theory of peer abuse. Sociological Studies of Children, 7, 177–205. Asher, S.R., & Wheeler, V.A. (1985). Children’s loneliness: A comparison of rejected and neglected peer status. Journal of Consulting and Clinical Psychology, 53, 500– 505. *Austin, S., & Joseph, S. (1996). Assessment of bully/victim problems in 8- to 11yearolds. British Journal of Educational Psychology, 66, 447–456. Baumeister, R.F., & Leary, M.R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497– 529. Besag, V.E. (1989). Bullies and victims in schools. Milton Keynes, U.K.: Open University. Birleson, P. (1981). The validity of depressive disorder in childhood and the development of a self rating scale: A research report. Journal of Child Psychology and Psychiatry, 22, 73–88. Björkqvist, K. (1994). Sex differences in physical, verbal, and indirect aggression: A review of recent research. Sex Roles, 30, 177–188. *Björkqvist, K., Ekman, K., & Lagerspetz, K. (1982). Bullies and victims: Their ego picture, ideal ego picture and normative ego picture. Scandinavian Journal of Psychology, 23, 307–313. *Boivin, M., & Hymel, S. (1997). Peer experiences and social self-perceptions: A sequential model. Developmental Psychology, 33, 135–145. *Boivin, M., Hymel, S., & Bukowski, W.M. (1995). The roles of social withdrawal, peer rejection, and victimization by peers in predicting loneliness and depressed mood in childhood. Development and Psychopathology, 7, 765–685. *Boulton, M.J., & Smith, P.K. (1994). Bully/victim problems in middle-school children: Stability, self-perceived competence, peer perceptions and peer acceptance. British Journal of Developmental Psychology, 12, 315–329. Boulton, M.J., & Underwood, K. (1992). Bully/victim problems among middle school children. British Journal of Educational Psychology, 62, 73–87. *Byrne, B.J. (1994). Bullies and victims in a school setting with reference to some Dublin schools. The Irish Journal of Psychology, 15, 574–586. 1
Data from the asterisked papers were included in the meta-analysis.
Annual progress in child psychiatry and child development 2000–2001
484
*Callaghan, S., & Joseph, S. (1995). Self-concept and peer victimization among schoolchildren. Personality and Individual Differences, 18, 161–163. Craig, W.M. (1998). The relationship among bullying, victimization, depression, anxiety, and aggression in elementary school children. Personality and Individual Differences, 24, 123–130. Crick, N.R., & Bigbee, M.A. (1998). Relational and overt victimization: A multiinformant approach. Journal of Consulting and Clinical Psychology, 66, 337– 347. *Crick, N.R., & Grotpeter, J.K. (1996). Children’s treatment by peers: Victims of relational and overt aggression. Development and Psychopathology, 8, 367–380. Crick, N.R., Wellman, N.E., Casas, J.F., O’Brien, K.M., Nelson, D.A., Grotpeter, J. K., & Markon, K. (1999). Childhood aggression and gender: A new look at an old problem. In D.Bernstein (Ed.), Nebraska Symposium on Motivation, 45. Lincoln, NE: University of Nebraska Press. Cunningham, C.E., Cunningham, L.J., Martorelli, V., Tran, A., Young, J., & Zacharias, R. (1998). The effects of primary division student-mediated conflict resolution programs on playground aggression. Journal of Child Psychology and Psychiatry, 39, 653–662. Dawkins, J. (1995). Bullying in schools: Doctors’ responsibilities. British Medical Journal , 310, 274–275. Eason, L.J., Finch, A.J., Jr., Brasted, W., & Saylor, C.F. (1985). The assessment of depression and anxiety in hospitalized pediatric patients. Child Psychiatry and Human Development, 16, 57–64. Farrington, D.P. (1993). Understanding and preventing bullying. In M.Tonry (Ed.), Crime and Justice, Vol. 17. Chicago: University of Chicago. Gilbert, P. (1992). Depression: The evolution of powerlessness. Hove, U.K.: Erlbaum. Graham, S., & Juvonen, J. (1998). Self-blame and peer victimization in middle school: An attributional analysis. Developmental Psychology, 34, 587–599. Hammen, C.L., & Glass, D.R. (1975). Depression, activity and evaluation of reinforcement. Journal of Abnormal Psychology, 84, 718–721. Harter, S. (1985). Manual for the Self-Perception Profile for Children. Denver, CO: University of Denver. Haselager, G.J.T. (1997). Classmates: Studies on the development of their relationships and personality in middle childhood. Unpublished doctoral thesis, Catholic University of Nijmegen, The Netherlands. Hawker, D.S.J., & Boulton, M. J. (1998). Peer victimization: Cause and consequence of psychosocial maladjustment? Manuscript submitted for publication. Hazier, R.J., & Hoover, J.H. (Eds.). (1996). Putting bullies out of business. [Special Issue]. Journal of Emotional and Behavioural Problems, 5(1). Howell, D.C. (1992). Statistical methods for psychology (3rd ed.). Belmont, CA: Wadsworth. *Kochenderfer, B.J., & Ladd, G.W. (1996). Peer victimisation: Cause or consequence of school maladjustment? Child Development, 67, 1305–1317. Kovacs, M. (1992). Children’s Depression Inventory Manual. North Tonawanda, NY: Multi-Health Systems.
Twenty Years’ research on peer victimization and psychosocial maladjustment
485
Kumpulainen, K., Raesaenen, E., Henttonen, I., Almqvist, F., Kresanov, K., Linna, S. L., Moilanen, I., Piha, J., Puura, K., & Tamminen, T. (1998). Bullying and psychiatric symptoms among elementary school-age children. Child Abuse and Neglect, 22, 705– 717. Kupersmidt, J.B., & Patterson, C.J. (1991). Childhood peer rejection, aggression, withdrawal, and perceived competence as predictors of self-reported behavior problems in preadolescence. Journal of Abnormal Child Psychology, 19, 427–449. Kupersmidt, J.B., Voegler, M.E., Sigda, K.B., & Sedikides, C. (1997, April). Social self discrepancy theory: Social self-cognitions as mediators of peer relations problems and concurrent adjustment. In A.H.N.Cillessen & M.E.DeRosier (Chairs), Linking peer relations and adjustment: The role of interpersonal perceptions. Symposium conducted at the Biennial Meeting of the Society for Research in Child Development, Washington, DC. *Lagerspetz, K.M.J., Björkqvist, K., Berts, M., & King, E. (1982). Group aggression among school children in three schools. Scandinavian Journal of Psychology, 23, 45– 52. Leary, M.R. (1990). Responses to social exclusion: Social anxiety, jealousy, loneliness, depression, and low self-esteem. Journal of Social and Clinical Psychology, 9, 221– 229. MacLeod, M., & Morris, M. (1996). Why me? Children talk to ChildLine about bullying. London: ChildLine. *Mynard, H., & Joseph, S. (1997). Bully/victim problems and their association with Eysenck’s personality dimensions in 8- to 13-year-olds. British Journal of Educational Psychology, 67, 51–54. *Neary, A., & Joseph, S. (1994). Peer victimization and its relationship to self-con-cept and depression among school-girls. Personality and Individual Differences, 16, 183– 186. *Olweus, D. (1978). Aggression in the schools: Bullies and whipping boys. Washington, DC: Hemisphere (Wiley). Olweus, D. (1993a). Bullying at school: What we know and what we can do. Oxford, U.K.: Blackwell. Olweus, D. (1993b). Victimization by peers: Antecedents and long-term consequences. In K.H.Rubin & J.B.Asendorpf (Eds.), Social withdrawal, inhibition and shyness in childhood. Hillsdale, NJ: Erlbaum. *O’Moore. A.M., & Hillery.B. (1991). What do teachers need to know. In M.Elliott (Ed.), Bullying: A practical guide for coping in schools. Harlow, U.K.: Longman. Parkhurst, J.T., & Asher, S.R. (1992). Peer rejection in middle school: Subgroup differences in behavior, loneliness, and interpersonal concerns. Developmental Psychology, 28, 231–241. Perry, D.G., Kusel, S.J., & Perry, L.C. (1988). Victims of peer aggression. Developmental Psychology, 24, 807–814. Perry, D.G., Perry, L.C., & Kennedy, E. (1992). Conflict and the development of antisocial behaviour. In C.U.Shantz & W.W.Hartup (Eds.), Conflict in child and adolescent development. Cambridge, U.K.: Cambridge University Press. Pierce, S. (1990). The behavioral attributes of victimized children. Unpublished master’s
Annual progress in child psychiatry and child development 2000–2001
486
thesis, Florida Atlantic University. Randall, P. (1996). A community approach to bullying. Stoke-on-Trent, U.K.: Trentham Books. Rigby, K. (1996). Bullying in schools: And what to do about it. London: Jessica Kingsley. *Rigby, K. & Slee, P.T. (1992). Dimensions of interpersonal relation among Australian children and implications for psychological well-being. The Journal of Social Psychology, 133, 33–42. Rivers, I. (in press). The long-term consequences of bullying. In C.Neal & D.Davies (Eds.), Further client issues in pink therapy. Buckingham, U.K.: Open University Press. Rivers, I., & Smith, P.K. (1994). Types of bullying behaviour and their correlates. Aggressive Behavior, 20, 359–368. Rosenthal, R. (1984). Meta-analytic procedures for social research. Beverly Hills, CA: Sage. Rosenthal, R. (1995). Writing meta-analytic reviews. Psychological Bulletin, 118, 183– 192. Ross, D.M. (1996). Childhood bullying and teasing: What school personnel, other professionals, and parents can do. Alexandria, VA: American Counselling Association. Salmon, G., James, A., Cassidy, E.L., & Javaloyes, M.A. (1999). Bullying; a review: Clinical presentations to inpatient and outpatient psychiatric services. Manuscript submitted for publication. Salmon, G., James, A., & Smith, D.M. (1998). Bullying in schools: Self-reported anxiety, depression, and self- esteem in secondary school children. British Medical Journal, 317, 924–925. Schwartz, D., Dodge, K.A., & Coie, J.D. (1993). The emergence of chronic victimization in boys’ peer groups. Child Development, 64, 1755–1772. *Sharp, S. (1996). Self-esteem, response style and victimization: Possible ways of preventing victimiztion through parenting and school based training programs. School Psychology International, 17, 347–357. Skinner, A. (1996). Bullying: An annotated bibliography of literature and resources (2nd ed.). Leicester, U.K.: Youth Work Press. *Slee, P.T. (1994a). Life at school used to be so good. Youth Studies Australia, 1 (Summer), 20–23. *Slee, P.T. (1994b). Situational and interpersonal correlates of anxiety associated with peer victimisation. Child Psychiatry and Human Development, 25, 97–107. Slee, P.T. (1995a). Bullying in the playground: The impact of inter-personal violence on Australian children’s perceptions of their play environment. Children’s Environments, 12, 320–327. *Slee, P.T. (1995b). Bullying: Health concerns of Australian secondary school students. International Journal of Adolescence and Youth, 5, 215–224. *Slee, P.T. (1995c). Peer victimization and its relationship to depression among Australian primary school students. Personality and Individual Differences, 18, 57–62. Slee, P.T., & Rigby, K. (1993a). Australian school children’s self appraisal of
Twenty Years’ research on peer victimization and psychosocial maladjustment
487
interpersonal relations: The bullying experience. Child Psychiatry and Human Development, 23, 273–282. *Slee, P.T, & Rigby.K. (1993b). The relationship of Eyscnck personality factors and selfesteem to bully-victim behaviour in Australian schoolboys. Personality and Individual Differences, 14, 371–373. Slee, P.T., & Rigby, K. (1994). Peer victimisation at school. AECA Australian Journal of Early Childhood, 19, 3–10. Smith, P.K., & Sharp, S. (Eds.) (1994). School bullying: Insights and perspectives. London: Routledge. Stanley, L., & Arora, T. (1998). Social exclusion amongst adolescent girls: Their selfesteem and coping strategies. Educational Psychology in Practice, 14, 94–100. Strauss, C.C., Lahey, B.B., Frick, P., Frame, C.L., & Hynd, G.W. (1988). Peer social status of children with anxiety disorders. Journal of Consulting and Clinical Psychology, 56, 137–141. Strube, M.J. (1985). Combining and comparing significance levels from nonindependent hypothesis tests. Psychological Bulletin, 97, 334–341. Tattum, D., & Tattum, E. (1992). Bullying: A whole-school response. In N.Jones & E. B.Jones (Eds.), Learning to behave: Curriculum and whole-school management approaches to discipline. London: Kogan Page. *Vernberg, E.M. (1990). Psychological adjustment and experience with peers during early adolescence: Reciprocal, incidental, or unidirectional relationships? Journal of Abnormal Child Psychology, 18, 187–198. Vosk, B., Forehand, R., Parker, J.B., & Rickard, K. (1982). A multimethod comparison of popular and unpopular children. Developmental Psychology, 18, 571–575. Walker, L.S., & Greene, J.W. (1986). The social-context of adolescent self-esteem. Journal of Youth and Adolescence, 15, 315–322. West, D.A., Kellner, R., & Moorewest, M. (1986). The effects of loneliness—a review of the literature. Comprehensive Psychiatry, 27, 351–363. Whitney, I., & Smith, P.K. (1993). A survey of the nature and extent of bullying in junior/middle and secondary schools. Educational Research, 35, 3–25. Williams, K., Chambers, M., Logan, S., & Robinson, D. (1996). Association of common health symptoms with bullying in primary school children. British Medical Journal, 313, 17–19.
PART VI: SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
27 Charting the Relationship Trajectories of Aggressive, Withdrawn, and Agressive/Withdrawn Children During Early Grade School Gary W.Ladd and Kim B.Burgess
The premises examined in this longitudinal investigation were that specific behavioral characteristics place children at risk for relationship maladjustment in school environments, and that multiple behavioral risks predispose children to the most severe and prolonged difficulties. Aggressive, withdrawn, and aggressive/withdrawn children were compared to normative and matched control groups on teacher and peer relationship attributes, loneliness, and social satisfaction from kindergarten (M age=5 years, 7 months; N=250) through grade 2 (M age=8, 1; N=242). Children’s withdrawn behavior was neither highly stable nor predictive of relational difficulties, as their trajectories resembled the norm except for initially less close and more dependent relationships with teachers. Aggressive behavior was fairly stable, and associated with early-emerging, sustained difficulties including low peer acceptance and conflictual teacher-child relationships. Aggressive/ withdrawn children evidenced the most difficulty: Compared to children in the normative group, they were consistently more lonely, dissatisfied, friendless, disliked, victimized, and likely to have maladaptive teacher-child relationships. Findings are discussed with respect to recent developments in two prominent literatures: children at-risk and early relationship development.
INTRODUCTION Many developmentalists have argued that both the quality and quantity of children’s peer contacts promote interpersonal cooperation, negotiation, and mutuality, and these experiences are essential for the development of healthy relationships (Asher & Coie, 1990; Berndt & Ladd, 1989; Bukowski, Newcomb, & Hartup, 1996; Piaget, 1932/1965; Sullivan, 1953). When the quality or quantity of social interaction is compromised, however, as may be the case for withdrawn or aggressive children, there is a greater
Charting the relationship trajectories of aggressive
489
likelihood that children’s psychological and social development will be altered or impaired. Researchers have typically defined childhood aggression in terms of “confrontive” forms (e.g., physical acts, such as hitting and kicking, and verbal acts, such as arguing and teasing; Coie & Kupersmidt, 1983; Dodge, 1983; Ladd & Price, 1987; Quay & Werry, 1986), and “nonconfrontive” forms (e.g., indirect acts, such as harming a peer’s reputation or relationship; Cairns & Cairns, 1994; Crick & Grotpeter, 1995). In contrast, social withdrawal encompasses the constructs of shyness (Buss, 1986), behavioral inhibition (Kagan, Resnick, Clarke, Snidman, & Garcia-Coll, 1984), and passive withdrawal (Rubin & Asendorpf, 1993; Rubin & Mills, 1988). Although potential biopsychosocial causes of differing forms of withdrawal remain under investigation, researchers tend to agree that differing forms of withdrawal share a common manifestation: infrequent interaction with others. Recent evidence is consistent with the view that social withdrawal is a heterogeneous construct, and suggests that it may take different forms depending on the child’s age (Coplan, Rubin, Fox, Calkins, & Stewart, 1994; Ladd & Profilet, 1996; Younger & Daniels, 1992). Beyond the distinction between active isolation and passive withdrawal (Rubin & Mills, 1988; Younger & Daniels, 1992), however, research on subtypes of withdrawn children is at an early stage (Harrist, Zaia, Bates, Dodge, & Pettit, 1997), and there are little data on the validity of these distinctions, the continuity of proposed “subtypes” over time or development, and the relationship trajectories that may be associated with each profile. Also, little is known about the short- or long-term outcomes associated with comorbid profiles, such as children who display aggressive and withdrawn behaviors. Rather than attempting to draw further distinctions between types of aggressive or withdrawn children, the principal aim of this investigation was to focus on particular forms of aggression and withdrawal that are likely to pose risk for young children’s relationship development, and ascertain whether children who fit these behavioral risk profiles follow differential relationship trajectories from school entrance through the early primary school years. By comparing the trajectories of aggressive, withdrawn, and aggressive/withdrawn children within the same study, it was possible to determine whether differing behavioral styles, manifest early in children’s school careers, are linked to unique relationship outcomes with peers and teachers, or whether they are associated with similar outcomes in each relationship domain. On the one hand, specific behavior patterns might lead to problems in multiple relationship domains; on the other hand, the effects of such behaviors might be limited to specific types of relationships or relationship features. Children who exhibit confrontive forms of aggression were targeted be-cause many investigators have proposed that this type of aggression is a risk factor for early relationship development (for a review, see Coie, Dodge, & Kupersmidt, 1990). Essentially, children who engage in these behaviors tend to alienate others and discourage relationship development because they supply a higher ratio of “costs” to their interaction partners (e.g., partners directly experience aversive consequences; Berscheid & Walster, 1978; Kelley & Thibaut, 1978) relative to benefits. Also targeted were children who display a form of passive withdrawal (i.e., passive-asocial behavior) in the
Annual progress in child psychiatry and child development 2000–2001
490
school setting; these children interact infrequently because they withdraw from peers (as opposed to being isolated or rejected by peers), or choose to play alone. Over time, children who prefer not to interact, or who isolate themselves from social interaction, are less likely to learn social principles and skills that are essential to relationship formation and maintenance (e.g., reciprocity, coregulation of play, building of emotional ties; Hartup, 1983). There is evidence to suggest that aggressive and withdrawn children are at risk for adjustment problems, but less is known about the relational trajectories that develop for children with these behavioral risks as they enter grade school and progress through the primary grades. It has been shown with older samples, for example, that aggressive behaviors predict peer rejection, externalizing disorders, and academic failure (e.g., Dodge, 1983; Ledingham & Schwartzman, 1984; for a review, see Parker & Asher, 1987). A smaller body of evidence suggests that passive withdrawn behaviors predict internalizing problems, such as low self-esteem, anxiety, and depression in these same age groups (Hymel, Rubin, Rowden, & LeMare, 1990; Rubin, Hymel, & Mills, 1989). Thus, even though withdrawal may lead to negative outcomes just as aggression does, more is known about the consequences of aggression. Furthermore, more is known about aggression and withdrawal as correlates of maladjustment during middle childhood and adolescence than during early childhood; in particular, there is a dearth of information about the risks posed by withdrawal and aggression following major ecological transitions, such as school entrance. This is a liability because challenging periods such as school transitions may increase children’s vulnerability for maladjustment (Cairns & Cairns, 1994; Coie et al., 1993; Rutter, 1996). Even less is known about the risks posed by comorbid behavior patterns, such as those characterized by both aggression and withdrawal. The limited research conducted on aggressive/withdrawn children shows that they are likely to possess lower levels of perceived competence, and exhibit higher levels of academic failure, distractibility, dependency, and peer rejection (Hymel, Bowker, & Woody, 1993; Ledingham, 1981; Ledingham & Schwartzman, 1984; Milich & Landau, 1984). This pattern of behahavioral comorbidity has also been linked to social immaturity, including children’s use of ageinappropriate social skills (e.g., initiating contact by interrupting others, avoiding eye contact; Lyons, Serbin, & Marchessault, 1988). Thus, although aspects of children’s psychological adjustment have been examined, such as cognitive and behav-ioral correlates (e.g., problem solving, academic ability, social skills, attention to task, compliance; Pepler & Rubin, 1991), less attention has been focused on interpersonal outcomes, especially within the domain of peer and teacher relationships. With the exception of Ledingham and Schwartzman, even fewer studies have been conducted with young aggressive/withdrawn children, and results from their investigation were unclear because some of the Pupil Evaluation Inventory items used to tap withdrawal were confounded with other constructs (e.g., “someone who is too shy to make friends”). Also neglected have been the linkages between all three behavioral styles (i.e., aggression, withdrawal, and aggression/withdrawal) and young children’s subjective feelings and appraisals of their relationships, including loneliness and perceived relationship satisfaction. These omissions are surprising given that developmentalists have emphasized the need to understand how children develop relationships in the school
Charting the relationship trajectories of aggressive
491
setting. Sroufe and Rutter (1984), for example, assert that the development of healthy relationships and adjustment to school are essential tasks during early to middle childhood. Moreover, success in relationships with peers and teachers constitutes an important indicator of children’s adaptation to school, and there is evidence to suggest that relationship difficulties in school predict later psychopathology (Cowen, Pederson, Babigian, Izzo, & Trost, 1973; Parker & Asher, 1987; Roff, Sells, & Golden, 1972). The adjustment criteria examined in this investigation included three forms of peer relationships, including friendship, peer group acceptance, and peer victimization, as well as the subjective/affective experiences of loneliness and perceived relationship satisfaction. Also examined were features of children’s relationships with teachers, including closeness, conflict, and dependency. These features of the teacher-child relationship have been linked with young children’s school adjustment (Birch & Ladd, 1997; Pianta, Steinberg, & Rollins, 1995) and, thus, were included in this study to determine how behavioral risks may precede, and possibly affect, specific features of the teacher-child relationship. Collectively, the investigated features of children’s peer and teacher relationships were construed as indicators of children’s relational adjustment in the school setting. Specifically, having mutual friendships and being accepted by peers, as well as having close ties with teachers, were considered indicators of adaptive (healthy) relationship adjustment. Conversely, a lack of friendships and rejection and/or victimization by peers, as well as conflictual teacherchild relationships were considered markers of maladaptive relationship adjustment. To address these aims, we conducted a prospective longitudinal study in which the relationship adjustment of young withdrawn, aggressive, and aggressive/withdrawn children were examined from kindergarten through second grade. Assessments of children’s behaviors and relationships began soon after school entrance and were repeated across grade levels, so that it was pos-sible to: (a) identify children who exhibited the targeted risk behaviors in “new” settings (i.e., kindergarten classrooms), and examine the stability of these behaviors across grade levels; and (b) assess the relationships that children developed among previously unacquainted1 classmates and teachers, and the relationship trajectories that emerged across the primary years. To be specific, four groups of children were identified early in kindergarten—aggressive, withdrawn, aggressive/withdrawn (comorbid), and nonaggressive/nonwithdrawn (normative and matched controls)—and their relationship trajectories were charted across four points in time (i.e., fall and spring of kindergarten, grade 1, grade 2) on each of the following measures: classroom friendships, peer acceptance, peer victimization, loneliness, perceived peer relationship satisfaction, and teacher-child closeness, dependency, and conflict. Because aggressive and withdrawn behaviors vary in their salience and impact upon peers and teachers, it was hypothesized that children who exhibited these styles would follow differing relational trajectories in classroom settings. As argued, passive withdrawn behaviors are typically inoffensive and seldom impinge on others, whereas aggressive behaviors tend to be aversive and directed at others. Aggressive behaviors, therefore, are more likely to have detrimental effects on children’s peer and teacher relationships than are withdrawn behaviors, although it is possible that the self-isolating tendencies of withdrawn (i.e., passive-asocial) children may eventually cause social
Annual progress in child psychiatry and child development 2000–2001
492
learning and skill deficits that interfere with relationship formation and maintenance. Specific predictions for these risk groups were based on the premise that aggressive and withdrawn behaviors may affect relationship adjustment in either a concurrent or predictive manner via certain sequelae. Evidence indicating that aggression has rapid effects on children’s relationships (e.g., Dodge, 1983), and is a fairly stable characteristic (Cairns & Cairns, 1994; Moskowitz, Schwartzman, & Ledingham, 1985) led us to expect that aggressive children would be disliked by peers early in kindergarten and thereafter in grades 1 and 2. Aggressive children, because of their aversive behaviors, were also expected to have less close and more conflictual teacher-child relationships than nonaggressive children. By contrast, young withdrawn children were not expected to be rejected by peers or form conflictual teacher-child relationships because others are less likely to perceive their behaviors as aversive. However, because withdrawn children seldom initiate exchanges with peers and respond to peers’ initiations less often (Wanlass & Prinz, 1982), they were expected to have fewer classroom friendships. Withdrawn children may also act this way with teachers, resulting in less close teacher-child relationships. Further, if young withdrawn children have anxious/insecure attachment histories (see Rubin & Lollis, 1988), and act in ways that are less assertive and more submissive than agemates (Rubin & Asendorpf, 1993), they may form more dependent relationships with their teachers. The most severe and prolonged relationship difficulties were expected for children whose behavior patterns manifested a combination of aggression and withdrawal. This comorbid behavioral pattern may compound children’s relationship difficulties because: (a) both aspects of this behavior pattern are likely to foster relationship maladjustment (i.e., constitute multiple or combined risks; Rutter, 1980; Sameroff & Seifer, 1983); and (b) the combination of aggression and withdrawal may have different interpersonal meanings and consequences for children’s interaction partners (e.g., peers and teachers) than do either of the individual behaviors that are manifest within this pattern. Thus, the greater risk attributed to the comorbid behavior pattern may be attributable to differences that are both quantitative and qualitative in nature. First, children who possess the comorbid behavior pattern are likely to display a larger quantity and range of atypical social behaviors; for example, they exhibit high levels of aggression and high levels of withdrawal. Second, peers’ and teachers’ experiences with aggressive/withdrawn children may be qualitatively different from those they have with either aggressive or withdrawn children because comorbid children display behaviors that are indicative of both hostility toward others and self-isolating, distancing tendencies. Such a behavior pattern may convey a different social meaning than does aggression or withdrawal alone. Children who alternate between displays of aggression and withdrawal are likely to be viewed by others as inconsistent and unpredictable (e.g., untrustworthy), which may adversely affect relationship formation and maintenance. Thus, compared to children in the other risk groups, aggressive/withdrawn children were expected to make fewer friends, exhibit 1
Acquaintance between children and their kindergarten classmates was assessed via parent questionnaire; on average, only 14.2% of children were acquainted with one or more classmates prior to school entrance.
higher levels of peer rejection, and have less close and more conflictual teacher-child
Charting the relationship trajectories of aggressive
493
relationshps due to the combined risks associated with this behavioral pattern (e.g., being perceived by others as aversive, unavailable, unpredictable). Moreover, if aggressive/withdrawn children fail to establish a supportive social network, they might also be expected to feel higher levels of loneliness and social dissatisfaction. With respect to sequelae, it also was of interest to examine the point in time at which children in the three behavioral risk groups began to display maladaptive relationships with peers and/or teachers (e.g., early versus later in kindergarten, grades 1 or 2). One possibility is that the putative effects of children’s behavioral styles on peer and teacherchild relationships emerge early in children’s school careers (e.g., fall of kindergarten) and remain consistent over time, as was predicted for aggressive behavior and peer rejection. Another possibility is that the effects of behavioral risks accumulate over time and amplify negative relationship trajectories (i.e., evidence cumulative continuity; Caspi, Elder, & Bem, 1987, 1988). A third possibility is that behavioral risks have delayed effects on children’s relationships. Olson (1992), for example, found that aggressive preschoolers were not victimized by peers early in the school year, but were later in the year. She suggested that peers’ counterattacks may have been delayed until networks (peer groups) had solidified, and peers were able to gang up on aggressors. The data gathered in this investigation provided an opportunity to determine which of these possibilities (i.e., temporal patterns in trajectories) characterized the relationship development of aggressive, withdrawn, and aggressive/withdrawn children.
METHOD Participants Participants consisted of two cohorts of kindergarten children (N=399) and their classroom teachers (cohort 1: n=16; cohort 2: n=18). Both cohorts were part of a larger longitudinal project conducted in schools from several U.S. communities. The communities selected for this study were chosen to represent a variety of demographic characteristics and ranged from rural to moderately urban in size and location, and families came from diverse socioeconomic backgrounds: 36.8% were lower to middle income ($0 to $20,000); 30.6% were middle income ($21,000 to $40,000); and 32.6% were upper-middle to high income (above $41,000). Parents’ informed consent was obtained as children entered kindergarten, and 80% or higher permission rates were achieved in all classrooms. In the fall of kindergarten, four risk/control groups (n=250; M age=5 years, 7 months) were selected from the larger sample (selection criteria described in the following), and 242 children remained in this sample at grade 2 (M age=8, 1). Of the eight subjects that were lost, seven were from the nonaggressive/nonwithdrawn (normative control) group and one was from the aggressive group. The composition of the risk/control sample was European American (76.8%), African American (17.6%), and Hispanic American, mixed race, or other (5.6%).
Annual progress in child psychiatry and child development 2000–2001
494
Identification of Risk and Control Groups Teacher ratings of children’s peer-related aggressive and social behavior in the fall of kindergarten, obtained from the Aggressive with Peers and Asocial with Peers subscales (see Table 27.1) of the Child Behavior Scale (CBS; Ladd & Profilet, 1996), were used to assign children from the larger normative sample to one of four behavioral risk/control groups termed withdrawn, aggressive, aggressive/withdrawn, or nonaggressive/nonwithdrawn (normative control). Children classified as withdrawn (n=46; 22 males) scored above the 67th percentile on the Asocial subscale and below the 33rd percentile on the Aggression subscale. Children assigned to the aggressive group (n=61; 40 males) scored above the 67th percentile on the Aggression subscale and below the 33rd percentile on the Asocial subscale. Those assigned to the aggressive/withdrawn group (n=34; 24 males) scored above the 67th percentile on both the Aggression and Asocial subscales. A nonaggressive/nonwithdrawn (normative control) group (n=109; 50 males) was created by selecting children with scores that fell below the 33rd percentile on both the Aggression and Asocial subscales. Also, because children in the aggressive and comorbid groups differed from the normative controls on gender and status risk (e.g., low income), a matched control group was created (n=45; 31 males) to control for these potential confounds in analyses designed to examine group differences over time (details in Results). A one-factor analysis of variance (ANOVA) (risk/ control groups) performed on scores for children’s chronological age revealed no significant age differences, F(3, 245)=.89, ns. Evidence for the Validity of the Risk-Group Designations. For validity purposes, it was of interest to determine whether: (a) children assigned to each risk group behaved in ways that were, on average, consistent with their riskgroup designations; and (b) children in the comorbid group displayed aggressive and withdrawn behaviors that were comparable to those exhibited by children who were assigned to the aggressive-only and withdrawnonly groups, respectively. Evidence bearing on these aims came from research on the validity of the CBS subscales (see Ladd & Profilet, 1996), and from additional measures that were administered in the larger longitudinal study that tapped convergent/discriminant aspects of the investigated risk behaviors. Ladd and Profilet (1996) examined the reliability and validity of the CBS subscales, including those tapping aggression and multiple forms of withdrawn behavior (i.e., asocial, anxious-fearful, and isolation due to exclusion by peers), with two cohorts of kindergarten children. Analysis of CBS psychometric properties showed that all subscale scores were distinct, internally consistent, and stable over time. The construct validity of each subscale was examined by correlating scores with convergent and discriminant measures gathered from multiple sources (i.e., observers, teachers, peers) across multiple occasions, and the evidence produced within this validation paradigm provided substantial support (and across validation) for the validity of the aggression and asocial behavior subscales.
Charting the relationship trajectories of aggressive
495
Table 27.1 Items Included in the CBS Aggressive with Peers and Asocial with Peers Subscales
Subscale I. Aggressive with Peers Item Number and Abbreviated Stem 4. Fights 16. Bullies 23. Kicks, bites, hits 35. Aggressive 36. Taunts, teases 38. Threatens 48. Argues
Subscale II. Asocial with Peers Item Number and Abbreviated Stem 25. Prefers to play alone 31. Likes to be alone 32. Keeps peers at distance 51. Solitary child 55. Avoids peers 57. Withdraws from peer activities
Although limited in scope, additional data on the behaviors of children assigned to each of the risk and control groups was gathered from observers and teachers in the larger longitudinal study. Observers rated aggressive, asocial, and anxious-fearful behaviors (defined so as to be isomorphic with the corresponding CBS subscales) for a subsample of children assigned to the withdrawn, aggressive, aggressive/withdrawn, and nonaggressive/nonwithdrawn groups (i.e., ns=20, 33, 18, and 47, respectively) during the first 10 weeks of kindergarten—prior to the time at which the CBS was administered and used to identify the risk groups. Using procedures identical to those detailed in Ladd and Profilet (1996), observers used a 5-point scale to rate children on three occasions after directly observing their behaviors in classroom settings. Seventeen percent to 20% of observers’ ratings were conducted with a reliability judge, and scores for each construct were created by averaging the ratings across occasions (exact agreement on ratings [see Hetherington & Clingempeel, 1992] of aggression and associal behavior exceeded. 80; across occasions >.87). In addition to the CBS aggressive with peers and asocial subscales, teachers concurrently rated all children in the risk/control groups on the CBS anxiousfearful subscales, and on subsets of Teacher Report Form items (TRF; Achenbach, 1991) that corresponded to the forms of aggression and asocial behavior that were targeted in this study. For a confrontive aggression score, TRF ratings for items 3 (“Argues”), 37 (“Fights”), 57 (“Attacks”), and 97 (“Threatens”), were summed ( =.81), and for a passive-asocial score, items 42 (“Child would rather be alone than with others”) and 111 (“Withdrawn, doesn’t get involved with others”), were summed ( =.62). Planned comparisons were used to determine whether the risk groups differed from the normative controls in expected directions on the observational and teacher rating measures. The general pattern of results (see Table 27.2), with some exceptions, lent support to the validity of the risk-group designations. Compared to the normative control group: (a) children in the aggressive group received observer and teacher ratings that were significantly higher for aggression but not significantly different for asocial and anxious-fearful behaviors; (b) children in the with drawn group received observer and teacher ratings that were in the predicted directions (i.e., higher for asocial and anxiousfearful behaviors, but not for aggressive behavior), and all were significant except for the
Annual progress in child psychiatry and child development 2000–2001
496
difference obtained on observer-rated asocial behavior;2 >and (c) children in the 2
Note that differences in sample sizes (fewer children could be observed) provided greater power to detect group differences for the teacher-report measures than for the observational measures shown in Table 2.
Table 27.2 Means and Standard Deviations for Behavioral Risk Groups on Supplementary Observational and Teacher-Report Measures
Aggressive Behavior Groups
OBS-AG TRF-AG M(SD) M(SD)
Anxious-Fearful Behavior OBS-AF CBS-AF M(SD) M(SD)
OBSTRFASOC ASOC M(SD) M(SD) 1.24 (.43) .08 (.28) 1.11 (.29) 1.20 (.32)
1.90 (1.98) ↕ NC group .04 (.19) 1.44 (.74) .04 (.19) 1.07 (.19) 1.21 (.36) WD group .04 (.21) 1.61 (.84) .83 (.90) 1.40 (.78) 1.80 (.56) ↕ ↕ versus NC group 1.24 (.38) .04 (.19) 1.44 (.74) .04 (.19) 1.07 (.19) 1.21 (.36) CM group 1 61 2.53 1.82 (1.01) 1.18 (1.38) 1.15 (.29) 1.64 (.42) ↕ ↕ versus (1.00) (1.78) m ↕ m NC group 1.24 (.38) .04 (.19) 1.44 (.74) .04 (.19) 1.07 (.19) 1.21 (.36) Additional comparisons for the comorbid group CM group 1.61 2.53 1.82 (1.01) 1.18 (1.38) 1.15 (.29) 1.64 (.42) ↕ ↕ ↕ versus (1.00) (1.78) AG group 1.99 1.90 1.24 (.43) .08 (.28) 1.11 (.29) 1.20 (.32) (1.01) (1.98) CM Group 1.61 2.53 1.82 (1.01) 1.18 (1.38) 1.15 (.29) 1.64 (.42) versus (1.00) (1.78) ↕ ↕ WD group 1.06 (.18) .04 (.21) 1.61 (.84) .83 (.90) 1.40 (.78) 1.80 (.56) Note. Pairs of means (above/below each other) are significantly different, p<.05, when connected with a double-headed arrow (?), and approach a significant difference when separated by an m (i.e., marginal, p<.07); means without any symbol between them are not significantly different. AF=anxious-fearful, AG=aggressive, ASOC=passive-asocial, CBS=subscale scores from the Child Behavior Scale (Oct./Nov. of kindergarten), CD =comorbid, NC=normative control, OBS=observational rating (first 10 weeks of kindergarten), TRF=items summed from Teacher Reachert Form (Oct./Nov. of kindergarten), WD=withdrawn. AG groups versus
1.99 (1.01) ↕ 1.24 (.38) 1.06 (.18)
Asocial Behavior
Charting the relationship trajectories of aggressive
497
Table 2 show that children in the aggressive/withdrawn group resembled aggressive children in that they received similar ratings from observers and teachers for aggressive behavior, but differed from the aggressive group in that they were rated as significantly more asocial. Although children in the aggressive/withdrawn group were rated as sgnificantly more aggressive than those in the withdrawn group, they also resembled the withdrawn group in that they did not differ significantly from these children on observer and teacher ratings of asocial behavior. Overall, this pattern of evidence suggests that, on average, the children assigned to each risk group behaved in ways that were consistent with their risk group designations, and that the form of aggression and withdrawal that was characteristic of children in the comorbid group was similar to the behaviors that were manifested by children assigned to the aggressive-only and withdrawn-only risk groups, respectively. Moreover, the behavioral observation data suggest that children exhibited these behavior patterns early in kindergarten, prior to the assessment of peer and teacher relationships. Measures: Child Behaviors Child Behavior Scale (CBS). This teacher rating instrument taps six aspects of children’s classroom behavior, and all subscales possess acceptable reliability and validity (see Ladd & Profilet, 1996). Two CBS subscales were used in this study: the 7-item Aggression subscale contains both physical and verbal aspects of aggression ( Σ .89 at each time of measurement; items shown in Table 1), and the 6-item Asocial subscale taps passive-withdrawn behavior ( Σ .86 at each time of measurement; see Table 1). This aspect of withdrawal differs from others included in the CBS (e.g., Excluded by Peers— isolation stemming from peer rejection). Thus, the Asocial subscale used in this study reflects withdrawn behavior that is characteristic of the child rather than isolation that is imposed by peers. Measures: Peer and Teacher Relationships Friendship. Friendship nomination data were obtained during individual sociometric interviews with participants and classmates (those with informed parental consent). Children were shown pictures of their classmates and asked if they had a best friend in their class. Those who answered affirmatively iden-tified up to five best friends and, from this list, were asked to identify their very best friend. For each child, two friendship measures were created: (a) the number of mutual friendships was calculated by summing friendship nominations given by the child that were also reciprocated by the nominated peers (range= 0 to 5); and (b) participation in a very best friendship, which was scored 0 or 1 depending on whether the child’s very best friend nomination had been reciprocated (Parker & Asher, 1993). Peer Group Acceptance. A peer rating procedure (Asher, Singleton, Tinsley, & Hymel, 1979) was used to assess peer group acceptance. Children were trained to use a 3-point scale and then asked to sort pictures of their classmates according to how much they liked to play with each child at school. Faces were used to represent the three scale points: 2 (happy face=“a lot”), 1 (neutral face =“kind of), and 0 (sad face=“not much”). Peer
Annual progress in child psychiatry and child development 2000–2001
498
acceptance scores were created by averaging the ratings children received from classmates and standardizing them within classrooms. Peer Victimization. This construct was assessed with a 4-item self-report scale developed by Kochenderfer and Ladd (1996). During individual interviews children were asked to report the extent to which they had experienced four types of peer aggression, including: (a) direct verbal—having kids say mean things to them; (b) indirect verbal— having kids say bad things about them to other kids; (c) physical—being hit or kicked; and (d) general—being picked on. Responses were coded as 1=“no” (never), 2=“sometimes,” or 3= “a lot,” and scores were averaged to create a victimization measure ranging from 1 to 3. Scores obtained from this measure were reliable (∇ Σ .74 at all times of measurement), and have been shown to be valid with samples of young children (Kochenderfer & Ladd, 1997). Loneliness and Social Satisfaction. The Cassidy and Asher (1992) Loneliness and Social Dissatisfaction Questionnaire (LSDQ) was revised to create a measure of loneliness that was distinct from relationship satisfaction. Three LSDQ items that refer to loneliness and two others (i.e., “Is school a lonely place for you?” “Are you sad and alone at school?”) were combined into a lonelines subscale. Fourteen remaining LSDQ items (e.g., “Is it hard to get other kids at school to like you?”) were used to tap social satisfaction. Children rated each item on a 3-point scale: 1=“no,” 2=“sometimes,” and 3= “yes”; higher scores were indicative of greater loneliness or lesser social satisfaction. Subscale items loaded on separate factors, were internally consistent (∇ Σ .75 for both scales at all times of measurement), and were averaged to create separate loneliness and social satisfaction measures. Teacher-Child Relationship. Teachers rated 35 statements on the StudentTeacher Relationship Scale (STRS; Pianta, Steinberg, & Rollins, 1995) using a 1 (“definitely does not apply”) to 5 (“definitely applies”) scale. Consistent with Pianta’s findings, factor analyses applied to STRS at each assessment occasion (N=399) yielded three factors. In each case, items corresponding to the closeness, conflict, and dependency subscales loaded on their designated factors. These subscales were labeled: Closeness (11 items; e.g., “I share an affectionate, warm relationship with this child,” “This child openly shares his/ her feelings and experiences with me,” ∇ Σ .90 at all times of measurement); Dependency (4 items; e.g., “This child reacts strongly to separation from me,” “This child asks for my help when he/she does not really need help,” ∇ Σ .69); and Conflict (12 items; e.g., “This child and I always seem to be struggling with each other,” “Despite my best efforts, I am uncomfortable with how this child and I have gotten along,” ∇ Σ .93). Closeness, Dependency, and Conflict scores were created for each child by averaging item scores by subscale. Procedure Data were collected from children and teachers at four points in time: fall of kindergarten (Kf, October/November), spring of kindergarten (Ks, March/April), grade 1 (G1, March/April), and grade 2 (G2, March/April). Trained interviewers administered the sociometric, friendship, victimization, loneliness, and social satisfaction measures to each child in counterbalanced order during two 40minute sessions. Teachers completed the
Charting the relationship trajectories of aggressive
499
CBS and STRS at each time of assessment. Children received a small gift, and teachers were paid for their participation.
RESULTS Overview of Analytic Aims Statistical analyses were guided by three principal aims. First, continuity in the targeted behavioral risk patterns across grade levels was examined using correlational analyses and repeated measures ANOVAs. Because rank-order stability coefficients are sensitive to departures from normality (e.g., truncated range), coefficients were calculated not only for the sample of children (n= 250) who were assigned to the four risk/control groups, but also for the normative sample (N=399) from which risk groups were drawn. In addition, a series of repeated measures ANOVAs were used to examine continuity or change in the absolute (mean) levels of aggression or withdrawal exhibited by the children within each risk group. Second, the degree of independence versus convergence among the measures used to tap the three dimensions of children’s relational adjustment (i.e., children’s peer relations, loneliness and social satisfaction, and teacher-child relationships) was examined in correlational analy-ses. Finally, the relational trajectories for the four risk and control groups were examined with a series of repeated measures ANOVAs.3 These analyses were used to determine whether children assigned to the four behavioral risk/control groups differed (on average) over time on the measures used to tap relationship adjustment within each of the three interpersonal domains. Two sets of analyses were conducted: one in which children in the three risk groups (i.e., aggressive, withdrawn, comorbid) were compared to a normative control group, and another in which the risk groups were compared to a matched control group. Sex differences were not anlayzed because inclusion of this factor would have created small and disparate cell sizes and unreliable mean estimates for some groups. Females were underrepresented in the aggressive and comorbid groups, as has been typical in studies of confrontive forms of aggression. Nonconfrontive (e.g., covert, relational) aggression appears to become more common in middle childhood and adolescence (Cairns & Cairns, 1994), especially for females (Crick & Grotpeter, 1995). Differences attributable to sex were controlled in the RM-ANOVAs computed with the matched control group. 3
Growth curve analyses (hierarchical modeling) using HLM-4 (Bryk, Raudenbush, & Congdon, 1996) yielded group differences and trends that were similar to those reported for RM-ANOVAs. Because attrition was minimal and few data points were missing, findings for the RM-ANOVAs are reported.
Annual progress in child psychiatry and child development 2000–2001
500
Stability of the Targeted Risk Behaviors and Risk-Group Membership Continuity and Change in the Targeted Risk Behaviors. The stability coefficients calculated across assessment intervals (see Table 3) revealed that the scores children received for aggressive behavior were more stable over time (and teachers) than were those obtained for withdrawn behavior. These findings were relatively consistent across samples in that highly similar stability coefficients were obtained for both the larger normative sample, and the smaller risk control sample. To compare continuity versus change in the levels of aggressive and asocial behavior displayed by children in each of the four risk and control groups, two repeated measures ANOVAs were computed. These analyses were used to determine whether the average levels of aggressive and asocial behaviors, as reflected on the corresponding CBS subscales, differed for the risk and control groups over time. Separate 4 (Groups: aggressive, withdrawn, comorbid, nonaggressive/nonwithdrawn)×4 (Time: Kf; Ks; G1 G2) repeated measures analysis of variance (RM-ANOVAs) were computed for the aggressive and asocial behavior scores. The analysis that was performed on the aggression scores produced significant effects for Group, F(3, 240)=113.93, p<.001, Time, F(3, 720)=26.83, p<.001, and Group×Time, F(9, 720)=21.73, p< .001; as did the anlaysis performed on children’s asocial behavior scores: Group, F(3, 240)=64.52, p<.001, Time, F(3, 720)=13.79, p<.001, and Group× Time, F(9, 720)=17.15, p<.001. Main effects were further examined with post-hoc tests on means (Tukey’s HSD; p<.05), and interactions were dismantled with tests of simple effects followed by post hoc tests on means.
Table 27.3 Stability Coefficients for the Aggressive and Asocial with Peers Measures Across Assessment Intervals
Criterion
Normative Longitudinal Sample (N=399) Time of Measurement Ks G1 G2 Aggressive with peers (Kf) Kindergarten-fall (Ks) Kindergarten, spring (G1) Grade 1, spring (G2) Grade 2, spring Asocial with peers (Kf) Kindergarten, fall (Ks) Kindergarten, spring
.70b
.50b .59b
.43b .52b
Risk/Control Sample (n=250) Ks G1 G2 .78b
.56b .66**
.56b
.56b
.22b .17b
.52b .59b .64b
.18b .20b .15a
.58b
.14 .14
.26b .29b .16
Charting the relationship trajectories of aggressive
501
(G1) Grade 1, spring (G2) Grade 2, spring Note. G1=spring of grade 1; G2=spring of grade 2; Kf=fall of kindergarten; Ks=spring of kindergarten. ap<.05 bp<.01
p<.001. Results obtained for aggressive behavior (Figure 1A) showed that children in the aggressive and comorbid groups did not differ from each other,4 but received significantly higher aggression scores than did children assigned to either the withdrawn or normative control groups when these scores were averaged over time (Group effect), and also at each time of assessment (Group× Time interaction effect). Scores for aggressive behavior tended to decline over time (grade levels), such that the aggression scores that were averaged over groups (Time effect) were significantly lower in grades 1 and 2 than in either the fall or spring of kindergarten. Yet, this change was largely attributable to the downward shift in average aggression scores exhibited by children in the aggressive and comorbid groups during grades 1 and 2 (Group×Time effect). Although not as strong, the analysis performed on asocial behavior revealed similar trends for children who were assigned to the withdrawn and comorbid groups (Figure 1B). Scores for passive-asocial behavior were significantly higher for the withdrawn and comorbid groups as compared to the normative control and aggressive groups (Group effect), although dismantling of the Group ×Time interaction showed that this difference was significant at Kf and Ks, but not G1. By G2, however, children in the comorbid group (but not the withdrawn group) received scores for asocial behavior that were, again, significantly higher than those exhibited by children in the normative and aggressive groups. Children in the withdrawn and comorbid groups did not differ from each other at any point in time.5 4
These findings indicate that children in the comorbid group were not significantly more aggressive or more withdrawn than children in the aggressive and withdrawn groups, respectively.
5
These findings indicate that children in the comorbid group were not significantly more aggressive or more withdrawn than children in the aggressive and withdrawn groups, respectively.
Annual progress in child psychiatry and child development 2000–2001
502
Figure 27.1. Means, standard deviations, and plot of means for (A) aggressive and (B) withdrawn behavior for the four behavioral risk/control groups from kindergarten through grade 2. Relations Among the Interpersonal Adjustment Measures at Each Time of Assessment Correlations among the measures used within each of the three relationship adjustment domains (i.e., peer relations, loneliness and social satisfaction, teacher-child relationship) are shown concurrently, or for each time of assessment, in Table 27.4. The measures used
Charting the relationship trajectories of aggressive
503
to tap aspects of children’s peer relations correlated in expected directions: the peer acceptance and friendship measures were moderately positively correlated with each other, and were negatively albeit weakly correlated with peer victimization. Consistent with past evidence (e.g., Ladd & Coleman, 1997; Ladd, Kochenderfer, & Coleman, 1997), the peer acceptance and number of mutual friendships measures exhibited the highest level of convergence at each time of assessment. The magnitude of these relations (R2 R .38), however, did not contradict the view that each measure taps a separate construct. Analysis of the relation between children’s loneliness and perceived satisfaction with peer relationships produced moderate, negative correlations at each time of assessment. These findings are consistent with past evidence (Ladd et al., 1997) and the premise that young children’s feelings of peer alienation and satisfaction with the quality of their classroom peer relationships are separate, but inversely related constructs.
Table 27.4 Associations among Measures of Children’s Peer and Teacher Relationships: Concurrent Correlations for Each Time of Measurement
A. Peer Relations N of Mutual Friends Ver* Best Friendship Peer Victimization Kf Ks G1 G2 Kf Ks G1 G2 K f Ks G1 G 2 Peer acceptance .56 .62 .50 .57 .29 .21 .10 .17 −.18 −.15 −.18 −.07 Number of 46 .36 .26 .33 −.18 −.19 −.17 −.06 mutual friends Very best −.10 −.03 −.01 −.01 friendship B. Loneliness and Social Satisfaction Social Satisfaction Kf Ks G1 G2 Loneliness −.49 −.51 −.54 −.48 C. Teacher-Child Relationship Dependency Conflict Kf Ks G1 G2 Kf Ks G1 G2 Closeness .06 .11 −.05 .10 −.53 −.55 −.45 −.51 .38 .30 .53 .34 Dependency Conflict Note. N for all correlations=395; rs above 13 are signifiant at p<.05, and those above 17 are significant at p<.01. G1=spring of grade 1, G2=spring of grade 2, Kf=fall of kindergarten, Ks=spring of kindergarten. Relations among the teacher-child relationship dimensions were moderate in magnitude and consistent with those reported in past research (e.g., Birch & Ladd, 1997;
Annual progress in child psychiatry and child development 2000–2001
504
Pianta, Steinberg & Rollins, 1995). Conflict was negatively correlated with closeness, but positively correlated with dependency. Closeness and dependency were essentially unrelated. Relational Trajectories for the Risk/Control Groups from Kindergarten through Grade 2 A series of 4 (Group: aggressive, withdrawn, comorbid, normative control) ×4 (Time: Kf, Ks, G1, G2) RM-ANOVAs were computed to examine differences because of group, time, and groups over time in the scores obtained for each of the relational adjustment measures. Results for these analyses, including means, standard deviations, and a graphic depiction of trends, are shown in Figure 2. Children’s Relations with Classroom Peers. The analysis performed on children’s peer acceptance ratings produced a significant Group effect, F(3, 230)=28.41; p<.001; the remaining effects were not significant. Post hoc tests on means (Tukey’s HSD) revealed that children in the aggressive and comorbid groups received, on average, significantly lower sociometric ratings from classmates than did children in the withdrawn or normative groups. Moreover, peer acceptance ratings for children in the comorbid group were significantly lower than those received by children in all of the other risk/control groups, including aggressive children. This pattern of group differences showed considerable continuity over time: higher levels of disliking emerged for children in the aggressive and comorbid groups soon after school entrance (i.e., Kf), and this status was maintained across kindergarten and the next two grade levels (Figure 2A). Thus, children assigned to the aggressive group were consistently less well liked by peers from Kf to G2 and, compared to this group, children assigned to the comorbid group were even more disliked throughout the early grade school years. The analysis performed on the number of mutual friendships measure produced a significant Group effect, F (3, 230)=8.95; p<.001, and a Time effect that approached significance, F(9, 690)=2.59; p<.052. Tukey HSD tests showed that, averaging over time, children in the comorbid group possessed significantly fewer friends than did children in the aggresive, withdrawn, or normative groups (Figure 2B). The nearly significant Time effect reflects what may be a trend for all children to develop a larger network of friends as they progress through the early school years (Ladd, 1988). Participation in a very best friendship was relatively rare for all of the children in the four risk/control groups. The analysis performed on these scores yielded only a weak, but significant Time effect, F(9, 690)=1.57; p<.024. Post hoc tests revealed that, compared to the fall of kindergarten, children were less likely to possess a reciprocated very best friendship in grades 1 and 2 (Figure 2C).
Charting the relationship trajectories of aggressive
505
Figure 27.2. Means and standard deviations on the relational criteria for the behavioral risk/control groups at each time of assessment.
Annual progress in child psychiatry and child development 2000–2001
506
Charting the relationship trajectories of aggressive
507
Annual progress in child psychiatry and child development 2000–2001
508
The final analysis within this domain showed that children’s average peer victimization scores differed significantly by Group, F(3, 242)=4.77; p<.003, and Time, F(9, 726) =101.15; p<.0001. The Group×Time effect was not significant. Post hoc tests on the group means showed that children in the comorbid group had higher average peer victimization scores than did children in the withdrawn or normative group. Tests on means for the time effect showed that, averaging over groups, victimization scores declined significantly across the kindergarten year, and then increased significantly from the spring of kindergarten to the spring of grade 1, and from spring of grade 1 to spring of grade 2 (Figure 2D). Loneliness and Social Satisfaction. Singnificant Group, F(3, 242)=5.48; p <.001, and Time F(3, 726)=24.04; p<.001, effects were found in the analysis of children’s loneliness scores. Averaging over time, children in the comorbid group reported significantly higher levels of loneliness than did children in the withdrawn or normative groups, but did not differ from their counterparts in the aggressive group. Post hoctests on the Time effect showed that, although average levels of loneliness were relatively stable during kindergarten, they declined significantly from kindergaren levels by the spring of grade 1 and then increased significantly (relative to all prior levels) by the spring of grade 2 (Figure 27.2E). The analysis computed on children’s social satisfaction scores produced significant effects only for Group, F(3, 238)=3.61; p<.014, and Time F(3, 714)=4.77; p<.003. Post hoc comparisons showed that, compared to the normative group, only children in the comorbid group expressed significantly lower satisfaction with their classroom peer relations. All other group differences were nonsignificant. When averaged over group, means for the Time effect did not differ significantly across kindergarten and grade 1, but rose significantly over all prior levels by the spring of second grade (Figure 27.2F). Children’s Relations with Classroom Teachers. Without exception, the analyses performed on the three teacher-child relationship meansures yielded significant main and interaction effects. By measure, the F and p values for Group, Time, and Group X Time effects, respectively, were: Closeness, F(3, 240)= 28.67, p<.001; F(3, 720)=3.45, p<.016; F(−,720)=2.06 p<.031; Dependency, F(3, 240)=12.61, p<.001; F(3, 720)=19.68, p<.016; F(9, 720)= 3.98, p<.031; and Conflict, F(3, 240)=91.13, p<.001; F(3, 720)=43.59, p <.016; F(9, 720)=11.08. p< .031. Because interaction effects typically qualify the interpretaton of main effects, tests of simple effects followed by post hoc tests on means were used to dismantle each interaction term. Results for closeness (Figure 27.2G) showed that, compared to children in the normative group, children in the comorbid group received lower ratings for closeness at all times of assessment. In addtion, this feature of the teacher-child relationship was relatively stable for children in the comorbid group; closeness scores for this group did not change significantly from the fall of kindergarten through spring of grade 2. In contrast, closeness tended to increase over time for children in the withdrawn and aggressive groups, although this increment (i.e., Kf to G2) was significant only for children in the aggressive group. Comparisons between the aggressive and normative groups produced findings that paralleled those of the comorbid group. Averaged closeness ratings for children in the aggressive group were significantly lower than those for children in the normative group at all times of assessment except grade 2 (where these
Charting the relationship trajectories of aggressive
509
differences were nonsignificant). Withdrawn children received closeness ratings that were significantly lower than those assigned to children in the normative group only in the fall of kndergarten, and received closeness ratings that were significantly higher than those given to children in the comorbid group only during the spring of kindergarten. Follow-up analyses performed on dependency scores (Figure 27.2H) showed that, compared to children in the normative group, teachers rated their relationships with children in the comorbid group as more dependent at every time of assessment except grade 2. Withdrawn children also were rated as more dependent than children in the normative group, but only in the fall of kindergarten. Although the main effect of Time suggested that, ovrall, dependency declined as children progressed from kindergarten into the primary grades, trends in this direction were significant only for children in the normative, withdrawn, and comorbid groups. Once disassembled, the interaction obtained for the conflict dimension (Figure 27.2I) revealed that children in the aggressive and comorbid groups had significantly higher conflict scores than did children in the normative and withdrawn groups at every point in time. Comparisons between children in the aggressive and comorbid groups indicated that, in general, conflict scores did not differ significantly for these two groups, except during the spring of kindergarten, when the comorbid group received significantly higher conflict scores than did the aggressive group. Although analysis of the overall Time effect suggested that conflict declines as children progress into the primary grades, follow-up tests indicated that this temporal shift was significant only for children in the three risk groups. The absence of such a trend for the normative group may be attributable to a floor effect for this group on the conflict measure. Risk Group Trajectories Compared to Matched Controls Factors such as children’s gender, and aspects of their family backgrounds (i.e., family/child ethnicity, family income, and parents’ SES) have been implicated as potential “status risks” that may have an important bearing on children’s adjustment trajectories (Cairns & Cairns, 1994; Felner et al., 1995). To control for these potential confounds, a series of one-factor ANOVAs (Groups: aggressive, withdrawn, comorbid, normative controls) were used to determine whether children in the three behavioral risk groups differed from those in the normative group on each of the stauts risk dimensions. In each of these analyses, the main effect for Group was significant, F(3, 240)>.6.72, p <.01, but post hoc tests on means showed that the Withdrawn group did not differ from the normative group on any of the status risk dimensions. The aggressive and comorbid groups, however, did differ significantly from the normative group, p<.05, on gender composition (higher percentage of males: AG, 65.6%; CMB, 70.6% versus NC, 45.9%), on child ethnicity (higher percentage non-Caucasian: AG, 44%; CMB, 35.3% versus NC, 7.3%), on total family income (lower M income: AG, $26,000; CMB, $26,400 versus NC, $40,300), and on socioeconomic status (lower SES as measured by Entwisle and Astone’s [1994] SEI): AG, 40.94; CMB, 38.41 versus NC, 59.52). Therefore, a matched control group was created by selecting children from the normative control groups who matched the children in the aggressive and comorbid groups on all of these potential status risk variables. To confirm the relative equivalence of these groups, a one-factor
Annual progress in child psychiatry and child development 2000–2001
510
ANOVA (Groups: aggressive, withdrawn, comorbid, matched controls) was again calculated on each of the status risks measures, and results showed no significant group effect on any dimension. Next, the ANOVAs and post hocs used to analyze risk group trajectories in the previous section were recomputed after substituting the matched control group (n=45) for the normative control group. These analyses produced findings that replicated prior results with two exceptions. First, the means for the social satisfaction Group effect were in the same direction as reported for the prior analyses but not significantly different. Second, the Group×Time interaction for teacher-child closeness failed to achieve significance, and so the trend for aggressive children to become closer with their teachers over time also was nonsignificant. It is lkely that these results are attributable to the smaller sample size of the matched control group relative to the normative control group and, therefore, reduced power to detect differences.
DISCUSSION Consistent with the literature on early psychosocial risk, a guiding premise for this investigation was that certain child characteristics, in this case the social behaviors that children display as they enter formal schooling, interfere with their ability to adapt to the surrounding environment, and thus, constitute risks for dysfunction. This premise was examined with a sample of children drawn from community rather than clinic-referred populations, and data were gathered in the natural environment of early grade-school classrooms. Evidence of adaptation or dysfunction was obtained by comparing different risk groups in terms of their intial success with peer and teacher relationships, and their evolving trajectories within these interpersonal domains. Consequently, the obtained results make a number of important contributions to knowledge about at-risk populations and early relationship development and adjustment. First, evidence from this investigation addresses the question of whether aggressive and/or withdrawn behavior patterns can be construed as enduring behavioral “orientations” during the early grade school years. Person×Environment theories of adjustment (Coie et al., 1993; Garmezy, Masten, & Tellegen, 1984; Ladd, 1996) imply that it is possible to identify person characteristics that, either alone or in combination with environmental risks, increase the likeihood of adaptive difficulties and subsequent maladjustment. Among investigators who study young children, particular forms of behavior (e.g., ag-gression, social withdrawal) have commonly been viewed as potential risk factors, especially when these behaviors are displayed in interpersonal contexts (e.g., classrooms). Findings from this study, particularly those obtained by examining the stability of individual differences, tended to corroborate the contention that aggression is a stable child characteristic (e.g., see Ladd & Price, 1987; Moskowitz et al., 1985; Olweus, 1979): Children who were aggressive in kindergarten, soon after school entrance, also tended to be aggressive in grades 1 and 2. Moreover, it would appear that aggressive behavioral styles can be detected in classrooms at very early ages. Passiveasocial behavior was less stable than aggressive behavior, however, and these findings provide less support for the argument that this type of social withdrawal is an enduring
Charting the relationship trajectories of aggressive
511
behavioral orintation in young chidren.6 The lack of stability may partly account for why withdrawal was not associated with early relational maladjustment. Additional evidence about continuity and change in early behavioral risk came from the analysis of mean differences in aggressive and asocial behaviors, both by risk group and over time. Consistent with correlational findings, these analyses showed that children who were initially identified as aggressive or aggressive/withdrawn exhibited higher levels of aggression than their normative counterparts at every grade level. Aggressive behavior exhibited by children in these risk groups, however, tended to decline as they progressed from kindergarten into grades 1 and 2. This temporal trend is consistent with evidence indicating that children’s aggression tends to decline with age (Cairns & Cairns, 1994), and may be partly attributable to artifacts associated with extreme scores (e.g., regression to the mean). There also was a tendency for asocial behavior to decline with age among children who were initially identified as withdrawn (i.e., those in the withdrawn and aggressive/withdrawn groups). Unlike results for the aggressive group, however, it became more difficult to distinguish these groups from the normative sample as children progressed into the primary grades. As a second contribution, results from this investigation further elucidate the relation between risk behaviors and emerging adjustment difficulties in children’s relationships with peers and teachers. In particular, the reported findings shed light on whether aggression, withdrawal, and the combination of these behavior patterns operate as “risk factors” in the interpersonal environment of early grade school; that is, as behavioral markers that are associated with the onset and duration of children’s relational difficulties in this setting. Clearly, findings from this investigation provide substantial support for the inference that aggression, as exhibited by young children in new interpersonal settings (i.e., kindergarten classrooms), is associated with a maladaptive tra6
Although Ladd and Profile! (1996) found that teachers rated asocial behavior reliably over a school year, it is possible that changes in raters (across grades) contributed to this instability, especially if teachers are less accurate at identifying nonintrusive behaviors (as compared to aggression) at this age level.
jectory that is both early-emerging and enduring in both relationship domains. The findings allow a similar, if not stronger, case to be made for the relation between early comorbid behavior patterns, in this case aggression accompanied by withdrawal, and children’s relationship difficulties in the same two domains. Given the configuration of these trajectories relative to the norm—especially the tendency for adjustment difficulties to emerge early in kindergarten and remain relatively stable over grades (i.e., the nonescalating nature of maladjustment)—little support was obtained for a cumulative continuity or delayed effects interpretation. Only early withdrawal, particularly the propensity to engage in passiveasocial behavior, produced a pattern of findings that was inconsistent with a “risk” interpretation. Young withdrawn children did not appear to have substantial peer difficulties in kindergarten, nor did this style appear to place children on trajectories toward maladaptive peer relationships during the primary years. Contrary to expectation, withdrawn children had about as many mutual friends as did their normative counterparts. Although withdrawn children interact with peers less than average, they
Annual progress in child psychiatry and child development 2000–2001
512
may still interact occasionally and engage in parallel play; these encounters may be enough for withdrawn children to make or be considered a friend at this age. This form of passive withdrawal may have different outcomes at older ages, however, and other types of withdrawal (e.g., reticence) may yield different results. The fact that withdrawn children were not rejected by their peers at this age level was consistent with our hypothesis and corroborates prior evidence in the peer relations literature (Rubin & Asendorpf, 1993). Further, findings indicating that children with a propensity toward asocial behavior did not exhibit significant peer relational difficulties also may generalize to a second interpersonal domain—young children’s relationships with their classroom teachers. Examination of the teacher-child relationship showed that, although withdrawn children were significantly less close and more dependent on their teachers as they began school (i.e., compared to the norm), their subsequent trajectory on these relationship dimensions did not differ singnificantly from the normative sample. Apparently, children who begin school with a proclivity toward asocial behavior are slower than normal at forming mature, close ties with their teachers. This delay may stem from a preference for nonsocial over social activities, or from discomfort in less familiar environments. Such difficulties appear to be transient, however, or restricted to the period following school entrance. The fact that passive withdrawn children exhibited few lasting relational difficulties contradicted the premise that the propensity to play alone or engage in self-isolating behaviors confines children’s skill learning in ways that advesely affects their future relationships with peers and teachers. It may be the case that, during the early childhood years, this behavioral profile does not restrict children’s social learning, or cause negative relational consequences with peers and teachers. Or, as Harrist et al. (1997) have argued, even narrower subtypes of withdrawal may exist, and risk may be specific to only some forms of passive withdrawn behavior. These investigators distinguish between subtypes termed “unsociable” (i.e., children who prefer to play alone) and “passive-anxious” (i.e., children who avoid peers or play due to inhibiting emotions), and contend that unsociable children are likely to be more interpersonally skilled and therefore at less risk than passive-anxious children. This line of argument raises the possibility that we may have included some proportion of unsociable children in our passive withdrawn group and, therefore, reduced the probability of detecting relational difficulties for this risk group. It is important to note, however, that the relative absence of relational difficulties found for passive withdrawn children may not generalize beyond the primary grades. As investigators in the field of developmental psychopathology have argued (Cairns & Cairns, 1994; Coie et al., 1993; Sroufe & Rutter, 1984), factors that do not function as risks during the early stages of children’s lives may do so at later points in development, and early risks may have distal rather than proximal effects over the life cycle. The trajectory that emerged for passive withdrawn children in Rubin and colleagues’ longitudinal studies (e.g., Rubin, 1993; Rubin et al., 1989) appears to be consistent with these premises. These investigators found that, although withdrawn behavior was not associated with peer rejection during early childhood, it did correlate with peer rejection during middle childhood. One explanation for this shift in linkages is that, as passive withdrawn children remove themselves from the context of social interaction, skill deficits accrue and cause relational difficulties. Another hypothesis is that peers are less
Charting the relationship trajectories of aggressive
513
sensitive to withdrawal at early ages (Younger, Gentile, & Burgess, 1993) but begin to recognize these behaviors and judge them deviant as they grow older. Likewise, although the withdrawn children in this study were not lonely, withdrawal and loneliness appear to be correlated later in development as children approach middle childhood (Boivin, Hymel, & Bukowski, 1995; Hymel et al., 1990). Unlike withdrawal, aggression was associated with young children’s relational difficulties during the primary grades. In light of previous findings, it is not surprising that aggression emerged as a potential precursor of children’s relational problems in the classroom, particularly peer group rejection and conflictual teacher-child relationships. Aggression evidenced relatively high stability over time and raters (i.e., different teachers), suggesting that it consititutes a behavioral style that is maintained over substantial intervals during the early elementary years. Especially in transitional contexts, such as kindergarten, where children are relatively unacquainted with peers and teachers, such characteristics are likely to have a detrimental effect on early relationship formation and development, and activate negative relational trajectories (e.g., by confirming peers’ and teachers’ prior expectations, perpetuating reputational effects). One explanation for this effect, as argued here and elsewhere (see introduction to this article; Coie, 1990), is that peers find this class of behaviors aversive, and such experiences foster negative relational processes and outcomes, such as rejection and conflict. Such outcomes are well documented, especially in the peer domain (e.g., see Coie & Kupersmidt, 1983; Dodge, 1983; Ladd, Price, & Hart, 1988), and data from this investigation further corroborate this interpretation. Less has been known, however, about the effect that early aggressive styles may have on children’s relationships with others in the school setting, such as teachers. Our findings suggest that aggressive behavior patterns not only raise the probability that children will have difficulties with peers, but also increase the likelihood that they will develop conflictual relationships with teachers. These findings also extend prior knowledge by revealing that emergent relational difficulties often signal the onset of trajectories that will persist well into the primary grades. More novel and intriguing findings emerged for children whose behavior pattern during the fall of kindergarten was characterized by both aggression and withdrawal. Compared to research conducted on either aggression or withdrawal, this comorbid behavioral profile and its relational correlates have received far less empirical attention at this age level, and rarely have children who fit this profile been identified with instruments that tap a specific aspect of withdrawal (i.e., passive-asocial behavior) without confounding it with other constructs (e.g., active isolation). The relational prognosis for children who fit this profile was not a positive one and, in many respects, was more severe and enduring than the one that emerged for aggressive children. Compared to the normative sample, children in the comorbid group evidenced a broader range of relational difficulties soon after school entrance; specifically, they were less liked by classmates, had fewer reciprocated friendships, suffered higher levels of peer abuse (i.e., victimization), and were more distressed (i.e., lonely, dissatisfied) with their classroom peer relations. They also were more likely to form maladaptive relationships with teachers—specifically, ties that were less close, more conflictual, and more dependent. When examined over time, this pattern of relational maladjustment was quite stable: Children in the comorbid group evidenced very little recovery or improvement
Annual progress in child psychiatry and child development 2000–2001
514
during the primary grades. Exceptions to this overall trajectory were found for two dimensions of the teacher-child relationship: both teacher-child conflict and dependency declined significantly for aggressive/withdrawn children as they moved from kindergarten through grades 1 and/r 2. Even more disturbing, however, was the fact that children in the comorbid group were significantly more maladjusted than children in the aggressive group on two peer relationship dimensions. Beginning in kindergarten, and across the primary grades, aggressive/withdrawn children had fewer mutual friends and lower peer acceptance than did aggressive children. The fact that aggressive/withdrawn children evidenced a pattern of relationship difficulties that was more diverse and enduring than that observed for the other risk groups may be attributed to several factors. First, these results are consistent with the proposition that a combination of risk factors (i.e., multiple risks) increases children’s vulnerability for disorder, either additively or contingently (Coie et al., 1993; Rutter, 1980; Sameroff, 1996). Second, given that the comorbid group’s relational difficulties were more severe than those of the aggressive group, and withdrawal alone was not associated with relational adjustment difficulties, the effects of the combined risk factors may be best interpreted as evidence of a moderator effect. Withdrawal, when displayed in the presence of aggression, may take on a different meaning for peers and teachers than when children display this risk behavior alone. It may be that when peers and teachers contend with a confrontive-aggressive behavioral style, they tend to interpret accompanying withdrawn behaviors not as evidence of interpersonal anxiety/fearfulness, disinterest, or detachment, but rather as a sign of arrogance or contempt for others. Such interpretations may make aggressive/withdrawn children seem not only unpredictable in their actions, but also unattractive as interaction and relationship partners for both agemates and adults. Third, it is also possible that the withdrawn component of the aggressive/withdrawn profile is a marker for other types of social skill deficits. For example, aggressive/withdrawn children may fail to compensate for their aggressiveness the way some aggressive children do because they lack prosocial skills. Fourth, compared to aggressive children, there may be qualitative differences in the way aggressive/withdrawn children handle aggressive episodes. As Coie, Dodge, Terry, and Wright (1991) have suggested, some aggressive children (boys) have more intense and prolonged fights than do others (e.g., they escalate and persist until there is a clear winner or loser, rather than allowing the fight to end ambiguously). Such bouts are likely to create more severe enmities and prolonged relational difficulties. The results obtained in this investigation alos speak to other issues that have been raised in the literature on childhood risk and maladjustment. One important question is whether particular forms of behavioral risk increase children’s vulnerability for multiple types of adjustment problems or, alternatively, whether different types of adjustment problems have the same determinants (Coie et al., 1993). Evidence that reflects on this issue serves to illuminate the diversity of disorders that may be associated with specific forms of risk, and may assist in identifying forms of risk that are common versus unique to specific types of disorder. Because children’s adjustment was assessed in two different relational domains (peer and teacher), the data gathered in this study allowed us to examine the convergence of children’s trajectories in both interpersonal contexts. Aggressive behaviors were associated with both peer group rejection and conflictual
Charting the relationship trajectories of aggressive
515
teacher-child relationships throughout the primary years and, thus, may be a common antecedent of these adjustment problems. Aggression, it might be argued, constitutes a coercive-resistant style of coping that children apply to diverse social contingencies (e.g., peers, teachers)—essentially, a reliance on forceful tactics to influence or control other persons or events (e.g., instrumental forms), or to resist contingencies that are imposed by other persons or events (e.g., reactive forms). Use of these tactics may well underlie the type of relational problems that emerged for aggressive children with both peers and teachers. When interacting with peers and teachers, children’s use of coercive-resistant tactics is likely to supplant or subvert others’ aims and interests and, thereby, cause their partners to develop adversarial reactions (e.g., conflict, rejection) and negative emotional states (e.g., anger, disliking). It also is possible that the adjustment problems that aggressive/withdrawn children exhibited with peers and teachers have common antecedents. One possibility is that aggressive/withdrawn children rely on what might be termed a “hostile-avoidant” coping style in their dealings with classroom peers and teachers. The aggressive behaviors that aggressive/withdrawn children display may stem from a need to control social contingencies to obtain desired outcomes. In addition, these children may have strong aggressive-resistive reactions to peers’ overtures and provocations (e.g., attempts to influence them or their activities), and to teachers’ efforts to manage or control their classroom behavior. Aggressive/withdrawn children’s tendencies to play alone or distance themselves from others (the withdrawn component of this profile) may be a means of reducing social anxiety, or a way to limit “interference” from others. For example, by being socially uninvolved, aggressive/withdrawn children may reduce demands placed upon them by peers and teachers, and limit obligations that stem from relationships (e.g., responsibilities to partners, reciprocation of resources). In the peer context, such an aggressive/withdrawn pattern might produce many of the outcomes that were observed in this investigation. Children who are hostile toward others and seek to be alone are likely to distance themselves from peers, and have trouble making friends and earning peer group acceptance. By displaying a high level of aggressive reactivity and avoidant behaviors that may be interpreted as arrogance or contempt, these children also may become vulnerable to peer provocation and abuse (victimization) that will likely foster awareness of their own social difficulties (i.e., loneliness, social dissatisfaction; Kochenderfer & Ladd, 1996). Moreover, the tendency to respond to peer provocations with aggression may create a trajectory in which aggressive/ withdrawn children further distance themselves from peers and reduce opportunities to form satisfying peer relationships. This same coping style might produce the types of adjustment problems that aggressive/withdrawn children experienced with teachers. Whereas social withdrawal may elicit greater care and protection from the teacher (i.e., elements of dependency), aggressiveresistive reactions to the teachers’ control and management strategies may foster conflictual teacher-child relationships. In sum, results from this study address important issues that have been articulated in two prominent literatures within the field of child development, namely the investigation of children at-risk and research on children’s relationship development and adjustment. Consistent with propositions advanced in the literature on early psychosocial risk, children with specific behavioral styles (i.e., aggressive, aggressive/withdrawn) were
Annual progress in child psychiatry and child development 2000–2001
516
found to develop relational difficulties with peers and teachers early in kindergarten. Moreover, once established, these difficulties tended to coalesce into fairly stable trajectories that persisted throughout the primary grades. Children who exhibited multiple forms of behavioral risk (i.e., the aggressive/withdrawn pattern) tended to develop relational difficulties that were more severe and broader in scope than did children who manifested only one form of behavioral risk (i.e., aggression with peers). These findings are consistent with the premise that certain behavior patterns during childhood function as risks that increase the likelihood that children will develop interpersonal problems in the school environment. Results from this investigation also have important implications for investigators interested in prevention science, or the development of programs designed to modify early behavioral risks and/or promote adaptive relationships (Coie et al., 1993). Establishing and maintaining supportive interpersonal ties are essential adaptive tasks for children as they enter school, and their competence at these tasks constitutes not only an important indicator of adjustment during this transition, but also an important risk or protective factor for challenges that emerge during later periods of the life course.
ACKNOWLEDGMENTS The research presented in this article was supported by National Institute of Mental Health grants R01MH-49223(01) and R01MH-49223(02) to Gary W. Ladd. This article was partially prepared while the first author was a Fellow at the Center for Advanced Study in the Behavioral Sciences at Stanford, and with support provided by the Spencer Foundation (Grant 199400132). We are very grateful to the children, parents, and teachers who made this study possible, as well as to Pathways Project members for their assistance with data collection. ©1999 by the Society for Research in Child Development, Inc. All rights reserved. 0009–3920/99/7004–0009
ADDRESSES AND AFFILIATIONS Corresponding author: Gary Ladd, University of Illinois, 183 Children’s Research Center, 51 Gerty Dr., Champaign, IL 61820 USA; e-mail:
[email protected]. Kim Burgess is at the University of Maryland at College Park.
REFERENCES Achenbach, T. (1991). Manual for the Teacher’s Report Form and 1991 profile. Burlington: University of Vermont, Department of Psychiatry. Asher, S.R., & Coie, J.D. (1990). Peer rejection in childhood. New York: Cambridge University Press. Asher, S.R., Singleton, L.C., Tinsley, B.R., & Hymel, S. (1979). A reliable sociometric
Charting the relationship trajectories of aggressive
517
measure for preschool children. Developmental Psychology, 15, 443–444. Berndt, T.J., & Ladd, G.W. (1989). Peer relationships in child development. New York: Wiley. Berscheid, E., & Walster, E. (1978). Interpersonal attraction. Reading, MA: AddisonWesley. Birch, S.H., & Ladd, G.W. (1997). The teacher-child relationship and children’s early school adjustment. Journal of School Psychology, 35, 61–79. Boivin, M., Hymel, S., & Bukowski, W. (1995). The roles of social withdrawal, peer rejection, and victimization by peers in predicting loneliness and depressed mood in childhood. Development and Psychopathology, 7, 765–785. Bryk, A.S., Raudenbush, S.W., & Congdon, R.T. (1996). HLM: Hierarchical linear and nonlinear modeling with the HLM/2L and HLM/3L programs. Chicago: Scientific Software International. Bukowski, W.M., Newcomb, A.F., & Hartup, W.W. (1996). The company they keep: Friendship in childhood and adolescence. New York: Cambridge University Press. Buss, A.H. (1986). A theory of shyness. In W.Jones, J.Cheek, & S.Briggs (Eds.), Shyness: Perspectives on research and treatment (pp. 39–46). New York: Plenum Press. Cairns, R.B., & Cairns, B.D. (1994). Lifelines and risks: Pathways of youth in our time. Cambridge, UK: Cambridge University Press. Caspi, A., Elder, G.H., Jr., & Bem, D.J. (1987). Moving against the world: Lifecourse patterns of explosive children. Developmental Psychology, 23, 308–313. Caspi, A., Elder, G.H., Jr., & Bem, D.J. (1988). Moving away from the world: Lifecourse patterns of shy children. Developmental Psychology, 24, 824–831. Cassidy, J., & Asher, S.R. (1992). Loneliness and peer relations in young children. Child Development, 63, 350–365. Coie, J.D. (1990). Toward a theory of peer rejection. In S.R.Asher & J.D.Coie (Eds.), Peer rejection in childhood (pp. 365–401). New York: Cambridge University Press. Coie, J.D., Dodge, K.A., & Kupersmidt, J.B. (1990). Peer group behavior and social status. In S.R.Asher & J.D.Coie (Eds.), Peer rejection in childhood (pp. 17–59). New York: Cambridge University Press. Coie, J.D., Dodge, K.A., Terry, R., &. Wright, V. (1991). The role of aggression in peer relations: An analysis of aggressive episodes in boys’ play groups. Child Development, 62, 812–826. Coie, J.D., & Kupersmidt, J.B. (1983). Abehavioral analysis of emerging social status in boys’ groups. Child Development, 54, 1400–1416. Coie, J.D., Watt, N.F., West, S.G., Hawkins, D., Asarnow, J.R., Markman, H.J., Ramey, S.L., Shure, M.B., & Long, B. (1993). The science of prevention: A conceptual framework and some directions for a national research program. American Psychologist, 48, 1013–1022. Coplan, R.J., Rubin, K.H., Fox, N.A., Calkins, S.D., & Stewart, S.L. (1994). Being alone, playing alone, and acting alone: Distinguishing among reticence and passive and active solitude in young children. Child Development, 65, 129–137. Cowen, E. L., Pederson, A., Babigian, H., Izzo, L. D., & Trost, M. A. (1973). Long term follow-up of early detected vulnerable children. Journal of Consulting and Clinical
Annual progress in child psychiatry and child development 2000–2001
518
Psychology, 41, 438–446. Crick, N.R., & Grotpeter, J.K. (1995). Relational aggression, gender, and socialpsychological adjustment. Child Development, 66, 710–722. Dodge, K.A. (1983). Behavioral antecedents of peer status. Child Development, 54, 1386–1399. Entwisle, D.R., & Astone, N.M. (1994). Some practical guidelines for measuring youth’s race/ethnicity, and socioeconomic status. Child Development, 65, 1521–1540. Felner, R.D., Brand, S., DuBois, D.L., Adan, A.M., Mulhall, P.F., & Evans, E.G. (1995). Socioeconomic disadvantage, proximal environmental experiences, and socioemotional and academic adjustment in early adolescence: Investigation of a mediated effects model. Child Development, 66, 774–792. Garmezy, N., Masten, A.S., & Tellegen, A. (1984). The study of stress and competence in children: A building block for developmental psychopathology. Child Development, 55, 97–111. Harrist, A.W., Zaia, A.F., Bates, J.E., Dodge, K.A., & Pettit, G.S. (1997). Subtypes of social withdrawal in early childhood: Sociometric status and social-cognitive differences across four years. Child Development, 68, 278–294. Hartup, W.W. (1983). Peer relations. In E.M.Hetherington (Ed.), P.H.Mussen (Series Ed.), Handbook of child psychology: Vol. 4. Socialization, personality, and social development (pp. 103–196). New York: Wiley. Hetherington, E.M., & Clingempeel, W.G. (1992). Coping with marital transitions: A family systems perspective. Monographs of the Society for Research in Child Development, 57(2–3, Serial No. 227). Hymel, S., Bowker, A., & Woody, E. (1993). Aggressive versus withdrawn unpopular children: Variations in peer- and self-perceptions in multiple domains. Child Development, 64, 879–896. Hymel, S., Rubin, K.H., Rowden, L., & LeMare, L. (1990). Children’s peer relationships: Longitudinal prediction of internalizing and externalizing problems from middle to late childhood. Child Development, 61, 2004–2021. Kagan, J., Reznick, S., Clarke, C., Snidman, N., & Garcia-Coll, C. (1984). Behavioral inhibition to the unfamiliar. Child Development, 55, 2212–2225. Kelley, H.H., & Thibaut, J.W. (1978). Interpersonal relations: A theory of interdependence. New York: Wiley. Kochenderfer, B.J., & Ladd, G.W. (1996). Peer victimization: Cause or consequence of school maladjustment? Child Development, 67, 1305–1317. Kochenderfer, B.J., & Ladd, G.W. (1997). Victimized children’s responses to peers’ aggression: Behaviors associated with reduced versus continued victimization. Development and Psychopathology, 9, 59–73. Ladd, G.W. (1988). Friendship patterns and peer status during early and middle childhood. Journal of Developmental and Behavioral Pediatrics, 9, 229–238. Ladd, G.W. (1996). Shifting ecologies during the 5–7 year period: Predicting children’s adjustment during the transition to grade school. In A.Sameroff & M.Haith (Eds.), The five to seven year shift (pp. 363–386). Chicago: University of Chicago Press. Ladd, G.W., & Coleman, C.C. (1997). Children’s classroom peer relationship and early school attitudes: Concurrent and longitudinal associations. Early Education and
Charting the relationship trajectories of aggressive
519
Development, 8, 51–66. Ladd, G.W., Kochenderfer, B.J., & Coleman, C.C. (1997). Classroom peer acceptance, friendship, and victimization: Distinct relational systems that contribute uniquely to children’s school adjustment? Child Development, 68, 1181–1197. Ladd, G.W., & Price, J.M. (1987). Predicting children’s social and school adjustment following the transition from preschool to kindergarten. Child Development, 58, 1168– 1189. Ladd, G.W., Price, J.M., & Hart, C.H. (1988). Predicting preschoolers’ peer status from their playground behaviors. Child Development, 59, 986–992. Ladd, G.W., & Profilet, S.M. (1996). The Child Behavior Scale: A teacher-report measure of young children’s aggressive, withdrawn, and prosocial behaviors. Developmental Psychology, 32, 1008–1024. Ledingham, J.E. (1981). Developmental patterns of aggressive and withdrawn behavior in childhood: a possible method for identifying preschizophrenics. Journal of Abnormal Child Psychology, 9, 1–22. Ledingham, J.E., & Schwartzman, A.E. (1984). A 3-year follow-up of aggressive and withdrawn behavior in childhood: Preliminary findings. Journal of Abnormal Child Psychology, 12, 157–168. Lyons, J., Serbin, L.A., & Marchessault, K. (1988). The social behavior of peeridentified aggressive, withdrawn, and aggressive/withdrawn children. Journal of Abnormal Child Psychology, 16, 539–552. Milich, R., & Landau, S. (1984). A comparison of the social status and social behavior of aggressive and aggressive/withdrawn boys . Journal of Abnormal Child Psychology, 12, 277–288. Moskowitz, D.S., Schwartzman, A.E., & Ledingham, J.E. (1985). Stability and change in aggression and withdrawal in middle childhood and early adolescence. Journal of Abnormal Psychology, 94, 30–41. Olson, S.L. (1992). Development of conduct problems and peer rejection in preschool children: A social systems analysis. Journal of Abnormal Child Psychology, 20, 327– 350. Olweus, D. (1979). Stability of aggressive reaction patterns in males: A review. Psychological Bulletin, 86, 852–875. Parker, J.G., & Asher, S.R. (1987). Peer relations and later personal adjustment: Are lowaccepted children “at risk”? Psychological Bulletin, 102, 357–389. Parker, J.G., & Asher, S.R. (1993). Friendship and friendship quality in middle childhood: Links with peer group acceptance and feelings of loneliness and social dissatisfaction. Developmental Psychology, 29, 611–621. Pepler, D.J., & Rubin, K.H. (1991). The development and treatment of childhood aggression. Hillsdale, NJ: Erlbaum. Piaget, J. (1932/1965). The moral judgement of the child. New York: Free Press Pianta, R.C., Steinberg, M., & Rollins, K. (1995). The first two years of shcool: Teacherchild relationships and deflections in children’s classroom adjustment. Development and Psychopathology, 7, 295–312. Quay, H.C., & Werry, J.S. (1986). Psychopathological disorders of childhood. New York: Wiley.
Annual progress in child psychiatry and child development 2000–2001
520
Roff, M., Sells, S.B., & Golden, M.M. (1972). Social adjustment and personality development in children. Minneapolis: University of Minnesota Press. Rubin, K.H. (1993). The Waterloo Longitudinal Project: Correlates and consequences of social withdrawal from childhood to adolescence. In K.H.Rubin & J.B. Asendorpf (Eds.), Social withdrawal, inhibition, and shyness in childhood (pp. 291–314). Hillsdale, NJ: Erlbaum. Rubin, K.H., & Asendorpf, J. (1993). Social withdrawal, inhibition, and shyness in childhood. Hillsdale, NJ: Erlbaum. Rubin, K.H., Hymel, S., & Mills, R.S. (1989). Sociability and social withdrawal in childhood: stability and outcomes. Journal of Personality, 57, 237–255. Rubin, K.H., & Lollis, S.P. (1988). Origins and consequences of social withdrawal. In J.Belsky & T.Nezworski (Eds.), Clinical implications of attachment (pp. 219–252). Hillsdale, NJ: Erlbaum. Rubin, K.H., & Mills, R.S.L. (1988). The many faces of social isolation in childhood. Journal of Consulting and Clinical Psychology, 56, 916–924. Rutter, M. (1980). Changing youth in a changing society. Cambridge, MA: Harvard University Press. Sameroff, A.J. (1996, Fall). Democratic and republican models of development: Paradigms or perspectives. Developmental Psychology Newsletter, APA Division 7, American Psychological Association. Sameroff, A.J., & Seifer, R. (1983). Familial risk and child competence. Child Development, 54, 1254–1268. Sroufe, L.A., & Rutter, M. (1984). The domain of developmental psychopathology. Child Development, 55, 17–29. Sullivan, H.S. (1953). The interpersonal theory of psychiatry. New York: W.W.Norton. Wanlass, R.L., & Prinz, R.J. (1982). Methodological issues in conceptualizing and treating childhood social isolation. Psychological Bulletin, 92, 39–55. Younger, A.J., & Daniels, T.M. (1992). Children’s reasons for nominating their peers as withdrawn: passive withdrawal versus active isolation. Developmental Psychology, 28, 955–960. Younger, A.J., Gentile, C., & Burgess, K.B. (1993). Children’s perceptions of social withdrawal: changes across age. In K.Rubin & J.Asendorpf (Eds.), Social withdrawal, inhibition, and shyness in childhood (pp. 215–235). Hillsdale, NJ: Erlbaum.
PART VI: SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
28 Violent Behavior in Children and Youth: Preventive Intervention from a Psychiatric Perspective Group for the Advancement of Psychiatry, Committee on Preventive Psychiatry
Objective: To outline causative factors for the epidemic of violence among children and youth in North America and suggest roles for child and adolescent psychiatry in preventive intervention. Method: The committee used literature searches to identify biological, psychological, and sociocultural factors associated with violent behavior. Results: Children and youth are both victims and perpetrators of violence. Risk factors include socioeconomic status, difficult temperament, chronic illness, psychiatric comorbidity, and parental psychopathology. Access to firearms in a culture of violence presents a particularly serious risk. Protective factors include intact family structures, prosocial peer groups, and supportive communities. Preventive interventions include the following: universal, addressed to total population groups; selective, for at-risk populations; and indicated, for children and youth developing violent behavior. Universal interventions including gun control and improved perinatal care are helpful, and selective interventions such as gun-free zones around schools may be successful. Indicated programs such as gun confiscation and conflict resolution for youth at serious risk may be useful, but only when embedded within well-funded, clinically based, and communityfocussed programs. Single-emphasis programs such as “Boot Camps” have intuitive appeal, but their utility is doubtful. Conclusions: Violent behavior can be prevented, and child and adolescent psychiatrists must be more active in community preventive interventions. (J Am Acad Child Adolesc Psychiatry, 1999, 38(3):235– 241.) Key Words: violence, conduct disorders, risk factors, protective factors, preventive interventions.
Annual progress in child psychiatry and child development 2000–2001
522
INTRODUCTION There is general concern with juvenile violence in America. The magnitude of the problem, along with the increasingly severe nature of violence in youth (Centers for Disease Control and Prevention [CDC], 1994, 1995), overwhelms traditional treatment resources (Elliott, 1994). Direct treatment efforts reach relatively few children and their families, are labor-intensive, and have been only moderately successful (Offord & Bennett, 1994). Preventive maneuvers and practices influencing broad social policies for neighborhoods, families, and children offer the best chance for reducing the incidence of violent conduct disorders and promoting healthy growth and development. Because violence and related psychopathology are slow in developing and present multiple psychiatric markers, child mental health practitioners are in a unique position to identify these problems and intervene early. Preventive interventions aim to reduce the number of risk factors and to increase the number of protective factors for the child, the parent-child environment, and the wider environment. In a tripartite model of the role of prevention in the mental health intervention spectrum (Fig. 28.1; Institute of Medicine, 1994), universal measures impact total population groups, selective measures are designed for at-risk populations, and indicated measures target children and families with problems leading to manifest disruptive disorders associated with violent behavior.
Figure 28.1. The mental health intervention spectrum for mental disorders. Reprinted with permission from Reducing Risks for Mental Disorders: Frontiers for Preventive Intervention Research. Copyright 1994 by the National Academy of Sciences. Courtesy of the National Academy Press, Washington, DC.
Violent behavior in children and youth
523
We consider a range of risk and protective factors that might modify violent behavior based on the available evidence, and we conclude with a call for action by community child and adolescent psychiatrists.
METHOD The committee conducted computerized searches of Medline and Paperchase databases through 1997. It defined violence as a sequence of aggressive actions with the child or adolescent as the direct perpetrator or victim. The following key words were used: infants, children, adolescent, homicide/murder, violence, and killing. Reports were also obtained from Federal Bureau of Investigation (FBI) Uniform Crime Reports, the CDC, the National Center for Injury Prevention and Control, the Bureau of Justice Statistics, the Pacific Center for Violence Prevention, selected RAND publications, and the Institute of Medicine Report on Reducing Risks for Mental Disorders (1994). The committee selected articles and reports deemed most relevant to the topic.
SCOPE AND MAGNITUDE OF THE PROBLEM In 1994, the FBI Uniform Crime Reports indicated that overall violent crimes rose 1% from 1993 to 1994. Juvenile arrests under age 18 increased by 3%, whereas adult arrests showed virtually no change. Arrests of persons younger than 18 years of age for murder and nonnegligent manslaughter totaled 2,982—an increase of almost 75% over the 1985 total (FBI, 1994). Likewise, the violent crime index for juveniles increased by 68% between 1988 and 1992 (Snyder & Sickmund, 1995). The U.S. Department of Justice has recently noted a decrease (9.2% in 1996), but even so adolescent homicide rates remain at historically high levels (J.A.Mercy, personal communication, 1997) and could increase with the anticipated 15% increase in the number of adolescents by 2000 (Fox, 1995). The problem is widespread across the country. As detailed by Richters and Martinez (1993), the 1980s witnessed an extraordinary increase in home and community violence across the United States. Osofsky (1995) surveyed the parents of school children in a New Orleans housing development; more than 90% of children and youth had witnessed violence (shooting, stabbing, rape, etc.), and more than 50% had been victims of violence. Bell and Jenkins (1993) reported that community violence and its effects were widespread among African-American children on Chicago’s southside, leading to feelings of victimization, growing uneasiness, and increased aggression among the children and a strong belief that the black community itself was being threatened. Child and adolescent violence is not limited to blighted urban neighborhoods; it extends into suburbia as well (Pitts & Steiner, 1994). Kachur et al. (1996) conducted a nationwide survey of violent deaths in schools in the United States during 1992; they tallied 105 deaths occurring in communities of all sizes across 25 states. Homicide was the predominant cause of death (81%), and a firearm was usually involved (77%). The literature review demonstrated a number of patterns based on age, gender, and race. Jason et al. (1983) reviewed FBI Uniform Crime Reports of child homicide age data
Annual progress in child psychiatry and child development 2000–2001
524
and found two clusters: children aged 3 and younger who were victims of intrafamilial violence and youths aged 12 and older who were victims of extrafamilial violence. A CDC report (1994) found that for 1978 through 1987, the annual homicide rates for young African-American males were four to five times that of young African-American females, five to eight times that of young white males, and 16 to 22 times that of young white females. These data have led to conclusions that violence is particularly an AfricanAmerican problem; however, there is good evidence that the relationship between violence and race is confounded by the strong relationship between minority race and poverty (Snyder & Sickmund, 1995). Minority overrepresentation for all types of crime has become progressively more marked over the past two decades (Pagan et al., 1987). From 1980 to 1990, the arrest rate for white males aged 15 to 24 hovered at 12 to 13 per 100,000, whereas the rate for African-American males in the same age range increased by a factor of 10 (Bazemore & McKean, 1993). Urban/rural and cultural factors have an effect on violence patterns. From 1993 to 1994, the U.S. crime rate rose 1% in the nation’s cities, rose 2% in suburban counties, and was unchanged in rural counties (FBI, 1994). Population density is correlated with the rates of homicide, rape, aggravated assault, and robbery. However, variation between a given city’s neighborhoods may be quite marked. In Washington, DC, for instance, a few census tracts accounted for a large proportion of its homicides. Gang membership and culture may also lead to more violent behavior. Ordog et al. (1993), studying rival Los Angeles gangs, found that 50% of gunshot victims were African-American and 50% were Hispanic. Seventy percent of all gunshot wounds were the result of driveby shootings. Lyon et al. (1992) studied a group of incarcerated white and Latino youth and found higher rates of criminal behavior among gang members than nongang members. The epidemic of youth homicide victims appears to be a distinctly American phenomenon. Fingerhut and Kleinman (1990) compared the homicide victim rates of young males (15 to 20 years old) in 22 developed countries during the period 1986 through 1987. The U.S. had an overall rate of 21.9 per 100,000. The next highest rate— that of Scotland—was far behind at 5.0 per 100,000. If the United States rate could be reduced to that of Scotland, more than 3,000 lives would be saved annually. Cost of Violence The most tragic cost of violence is life itself, but the monetary costs are an additional burden on the nation. The annual fiscal cost for medical treatment of injuries caused by firearms is approximately $4 billion, whereas the medical cost for domestic violence is approximately $44 million annually (Mandel et al., 1993). The relative cost for children and adolescents represents a significant percentage of this total. The National Association of Children’s Hospitals and Related Institutions reported that in 1991 the average bill for a child wounded by firearms was $14,434. The total cost of both property and violent crime in America was estimated by Business Week to be at least $425 billion (Mandel et al., 1993). Indirect costs, including property loss, urban decay, medical care, private protection, and criminal justice, were $255 billion more.
Violent behavior in children and youth
525
RISK AND PROTECTIVE FACTORS The complex problem of youth violence and its origins must be approached from an epidemiological perspective, considering both risk and protective factors. This model stands in strong contrast to the single-event hypotheses often used by clinicians. Risk factors for violent behavior include the following: the presence of violence in the home or neighborhood, alcohol abuse, involvement in the drug trade, gun possession, overt criminal activities, and association with older delinquent adolescents and/or adults. The Effect of Guns There is an added risk factor for U.S. youth: an unprecedented access to firearms. More U.S. teenagers die from gunshot wounds than from all natural causes combined. Firearmrelated mortality accounts for almost half of all deaths among African-American teenagers. Firearm homicides among 15- to 19-yearolds increased from 5.8 per 100,000 in 1985 to 18.1 per 100,00 in 1993—a 212% increase (National Center for Health Statistics, 1993). In 1985, the rate of firearm homicide among African-American males aged 15 to 19 years was 37.4 per 100,000, whereas among whites of the same age the rate was 5.0 per 100,000. During the next 8 years, the rate of firearm homicide among white adolescents more than doubled to 12.8 per 100,000, but the rate among African-American adolescents more than tripled to 131.5 per 100,000 (Fingerhut, 1993). Guns are now used in three quarters of teenage murders, a rate three times higher than several years ago (Fox, 1995). Studies note the increased lethality of firearms, as well as the easy availability of firearms to youth (Ash et al., 1996). Having a gun in the home increased the likelihood of homicide threefold, and suicide fivefold (Pacific Center for Violence Prevention, 1993). The availability of guns is also a crucial factor in the discrepancy noted in suicide rates between two comparable cities in the United States and Canada. In Seattle the rate of suicide by handguns was 5.7 times that in Vancouver. For youth aged 15 to 24, the Seattle rate was 10 times higher (Sloan et al., 1988). Guns often take on special meanings for children and youth, ranging from symbols of strength and manhood to protective agents against the fear of assault and death (Ash et al., 1996). Male joblessness, with resultant poverty and family disruption, may be a major causal factor leading to aggressive violence and gang membership, itself often associated with drugdealing and other antisocial criminal activities (Sampson, 1987). A culture of urban poverty, homelessness, and social disorganization produces maternal and child risk factors such as low birth weight, cognitive impairment, and child abuse/neglect, which in turn constitute risk factors for crime and violence in adolescence and young adulthood. Invariably, risk factors are additive and follow a developmental sequence. Domains of risk such as organic or temperamental difficulties, disrupted attachments, family adversity, inconsistent parenting, and problems in parent-child relationships predict the early onset of disruptive behavior disorders (American Academy of Child and Adolescent Psychiatry, 1997). Child maltreatment, which includes frank physical abuse, sexual abuse, neglect, and emotional abuse, is an important risk event
Annual progress in child psychiatry and child development 2000–2001
526
among youths who have committed murder (Garbarino, 1995). Lewis (1992) evaluated the relationship between child maltreatment and violent behavior. Nearly all violent adults appear to have been violent as juveniles, and she identified “intrinsically vulnerable children” with cognitive, psychiatric, or neurological impairments. Such neuropsychiatrically impaired children, by virtue of their hyperactivity and impulsivity, were more likely to receive abuse from adults in their family settings. However, the relative contribution of childhood trauma is difficult to gauge, because many children who are abused in childhood do not commit murder or other violent acts. Protective factors to reduce the chance that a child will develop conduct disorder include good intelligence; easy disposition; an ability to get along well with parents, siblings, teachers, and peers; an ability to do well in school; having friends; being competent in nonschool skill areas; and having a good relationship with at least one parent and/or other significant adults. A positive, warm bond between parent and child early in life may lead to more prosocial behavior. The support of other significant adults in the community, prosocial peer groups, and good schools fostering academic success, responsibility, and self-discipline are associated with diminished risk for conduct disorder (RaeGrant et al., 1989). Extrapolating from these findings to more violent environments, the presence of such prosocial persons or environments may be crucial. However, such protective factors may not be sufficient to offset the effects of hostile models and pervasively violent environments. More extensive coordination of these protective factors may require evolution into full preventive programs (Institute of Medicine, 1994). The development of such programs should also take into account that we have insufficient data to decide how protective factors operate in contradistinction to risk factors and whether they exact their influence cumulatively, additively, or in isolation (American Academy of Child and Adolescent Psychiatry, 1997). Understanding the causal chains involving risk and protective factors for conduct disorder and violence, as well as the extraordinary resilience shown by some individual children and adolescents in violent environments, raises fascinating and complex questions that require much more study.
PREVENTIVE INTERVENTIONS Of all the problems in child and adolescent psychiatry, violent behavior is the one most suited to prevention: It develops slowly, with risk factors gradually accumulating over many years before overt violent behavior emerges. This pattern presents clinicians with multiple opportunities to intervene. There are two recent extensive reviews of this topic (American Academy of Child and Adolescent Psychiatry, 1997; Institute of Medicine, 1994). The following examples are grouped according to scope (universal, selective, indicated) for the purpose of clarity, but these boundaries are somewhat artificial. For example, some programs instituted for selected at-risk populations demonstrate such success that they are subsequently implemented for entire populations. Universal preventive measures target entire populations. Many such measures (e.g., gun control) involve the promulgation of far-reaching policies and procedures, which in turn require legislative authorization and funding. Here, political support is essential.
Violent behavior in children and youth
527
Truly universal interventions involving total populations of children (such as seatbelt laws) are uncommon, but several current initiatives appear to have positive implications for the prevention of youth violence. These include widespread programs to enhance prenatal care, maternal/ infant care and nutrition, and family management for preschool children and parents. In an interesting national intervention campaign in Norway, Olweus (1991) found that a school program against aggressive behavior and bullying resulted in less bullying, less delinquency, and more attachment to school. This universal intervention shows how a particular problem can be targeted with positive results. Sometimes selective programs can approach universality. The School Development Program started by Dr. James Comer, a child psychiatrist, in the late 1960s has now spread to more than 500 schools nationwide. This program involves parents, teachers, and administrators in child development-centered social and educational programs. Follow-up studies have shown marked academic achievement, with decreases in dropout rates, serious behavior problems, and teacher turnover (Comer et al., 1996). Selective preventive measures target identified at-risk populations. A num-her of examples involving various age groups are given. A prenatal/early infancy project for mothers with economic deprivation, poor prenatal health, self-damaging behaviors, and poor family management gave rise to improved maternal diet, reduced smoking during pregnancy, fewer preterm deliveries, higher birth weight babies, and less subsequent child abuse (Olds et al., 1988). In randomized, controlled, prospective outcome studies in the preschool range, positive effects were shown for the children of families with multiple risk factors. Relevant outcomes included academic success, behavioral problems, parenting skills, family management problems, and arrest rates. Some of these effects were only apparent after several years of follow-up. The Houston Parent-Child Development Center Program (Johnson, 1990) for preschool children suffering economic deprivation, academic failure, early behavior problems, and poor family management practices led to fewer behavioral problems and better family management practices. The Perry Pre-School Program (Weikart et al., 1986) showed similar long-term results, including fewer behavioral problems and subsequent arrests. Such programs, including the Head Start program, may therefore help to prevent delinquency (Zigler, 1993). For grade school children, interpersonal cognitive problem-solving programs gave rise to better problem-solving skills and fewer behavior problems in children with economic deprivation, poor impulse control, and early behavioral problems. A Baltimore program encompassing more than 2,300 children included community preventive intervention, mastery learning, and the “Good Behavior Game” (Kellam & Rebok, 1992). Risk factors addressed included academic failure, aggressive behavior, poor concentration, shyness, and depressive symptoms, and effects of the intervention included a drop in both aggressive and shy behavior as well as better cognitive competence. In Seattle, a social development program for a similar group gave rise to comparable results (Hawkins et al., 1992). Such selective prevention programs provide an excellent basis for prosocial development and may serve as protective factors against later violence in youth. There have been a reasonable number of selective interventions for children at risk for conduct disorders (American Academy of Child and Adolescent Psychiatry, 1997). Individual programs such as school-based conflict resolution training programs (Ash et al., 1996), gun-free zones around schools, evening curfews, weekend and evening
Annual progress in child psychiatry and child development 2000–2001
528
recreation programs, summer camps, job and training programs for youth at risk, and community policing for young people at risk have all been tried, with some positive results. For adolescents, the Positive Youth Development Program (Caplan et al., 1992) addressed early onset of drug use, favorable attitudes toward drugs, and environmental risks. The outcomes were better coping skills, better conflict resolution and impulse control, and less alcohol abuse. The similar but larger Alcohol Education Project (Hansen & Graham, 1991), targeting adolescents with favorable attitudes toward alcohol abuse, found that fewer teenagers wanted to use alcohol and participants had increased knowledge of alcohol risks after the intervention. However, other selective interventions have variable effects. Gun buy-back programs (Mendel, 1995) indicate that an increase in the number of guns obtained is not paralleled by a decrease in crime. On the other hand, programs reducing the passage of firearms across state lines appear to have been effective (Weil & Knox, 1996). A further weakness in these programs is that, although they appear to work in the short term, the gains often erode quickly if the programs are not embedded in the total social structure (Kann et al., 1993). Indicated preventive programs for young persons at imminent risk for violent behavior have been a focus of mental health professionals—including child and adolescent psychiatrists—for many years. Individual interventions including psychotherapy have not been shown to be effective when used in isolation, but there is a substantial database supporting the effectiveness of broad familyand parent-based psychotherapeutic interventions in the grade school years to reduce violence and related psychopathology (Patterson & Narrett, 1990). The alarming rise in crime, with its cost implications, has also given rise to considerable enthusiasm (especially among some politicians) for more draconian approaches. The long-term success of single-event intrusive programs such as “Boot Camps” has not been demonstrated (Henggeler & Schoenwald, 1994). However, some multisystemic diversion programs aiming to deal with chronic offender youths before they become adjudicated delinquents have shown promising results (Borduin, 1999). The recent “Boston miracle” in which no teenager had been killed by gunfire in nearly 2 years (U.S. Department of Justice, 1996) illustrated vividly how a total, wellfunded, community approach can be effective. Deterrent actions included arrests of gang members, apprehension of gun dealers, and identification of youths violating parole. Supportive interventions included the provision of more than 100 part-time teenage counselors, the development of a basketball league limited to gang members, and the coordination of city-funded community centers and local churches. All of this cost a lot of money—$20 million from juvenile justice initiative funds—but the money was well spent. Moreover, the ultimate outcome may be cheaper in the long run. The Rand Corporation (Greenwood et al., 1996) reported that programs concentrating on crime prevention among young people were more cost effective in reducing serious crime than mandatory sentences for adult repeat offenders. In contrast, the investment in prisons stimulated by “Three Strikes” laws has diverted significant sums from preventive programs.
Violent behavior in children and youth
529
CLINICAL IMPLICATIONS AND A CALL FOR ACTION To be effective in dealing with the problem of violence and related psychopathology, a paradigm shift is required. Because exclusive individual clinical interventions for violent conduct disorders do not work, the child and adolescent psychiatrist must seek opportunities to be a leader or team member in well-organized and well-funded community prevention efforts. Such efforts may be used on an early intervention basis or in indicated programs against violent behavior (as in the Boston program). As the data presented in this article suggest, it is possible for us to be extremely effective. This change in clinical identity from individual to community child and adolescent psychiatrists may be difficult, but it is both important and exciting. In addition, it is necessary for us to be involved in further research efforts to understand and treat violent behavior in children and adolescents. Our training programs therefore need to focus more on this evolution of the community child and adolescent psychiatrist from an individual therapist into a member of an active community team, taking a full part in needed clinical and research activities. The recent decline in youth crime and arrest rates is real and significant, but it does not take into account incarceration effects and may not be reflective of any success from current programs. To extend this decline, further research investment is needed to help us better understand this problem, along with scientifically based clinical prevention programs offering real hope for communities across the country that are committed to preventing violence. The clinical and research rewards of this endeavor are exciting, and the needs of the children and adolescents at risk for and from violent behavior are extreme. It is time for us to take back the role from which we originally developed in the earliest days of child and adolescent psychiatry.
REFERENCES American Academy of Child and Adolescent Psychiatry (1997). Practice parameters for the assessment and treatment of children and adolescents with conduct disorder. J Am Acad Child Adolesc Psychiatry 36(suppl): 122S–139S. Ash P, Kellermann AL, Fuqua-Whitley D, Johnson D (1996). Gun acquisition and use by juvenile offenders. JAMA 275:1754–1758. Bazemore G, McKean J (1993). Developing health care services for incarcerated youth. In: Hard Times; Healing Hands, Thompson E, Farrow JA, eds. Arlington, VA: National Center for Education in National and Child Health, pp 99–120. Bell CC, Jenkins EJ (1993). Community violence and children on Chicago’s south side. Psychiatry 56:46–54. Borduin CM (1999). Multisystemic treatment of criminality and violence in adolescents. J Am Acad Child Adolesc Psychiatry 38:242–249. Caplan M, Weissberg RP, Grober JS, Sivo PJ, Grady K, Jacoby C (1992). Social
Annual progress in child psychiatry and child development 2000–2001
530
competence promotion with inner-city and suburban young adolescents: effects on social adjustment and alcohol use. J Consult Clin Psychol 60:56–63. Centers for Disease Control and Prevention (1994). Homicides among 15–19 year old males—United States 1963–1991. MMWR Morb Mortal Wkly Rep 43:725–727. Centers for Disease Control and Prevention (1995). Suicide among children, adolescents and young adults—United States 1980–1982. MMWR Morb Mortal Wkly Rep 44:289– 291. Comer JP, Haynes NM, Joyner ET (1996). Rallying the Whole Village: The Comer Process for Reforming Education. New York: Teachers College Press. Elliott DS (1994). Youth Violence: An Overview. Pamphlet F-693, Center for the Study and Prevention of Violence. Boulder: University of Colorado, March. Pagan J, Slaughter E, Hart E (1987). Blind justice: the impact of race on the juvenile justice procedure. Crime Delinquency 33:224–258. Federal Bureau of Investigation (1994). Uniform Crime Reports for the United States. Washington, DC: US Government Printing Office, p 11. Fingerhut LA (1993). Firearm Mortality Among Children, Youth, and Young Adults 1– 34 Years of Age: Trends and Current Status United States 1985–1990. Hyattsville, MD: National Center for Health Statistics. Advance Data for Vital Statistics. No. 231:1–20. Fingerhut LA, Kleinman JC (1990). International and interstate comparisons among young males. JAMA 263:3292–3295. Fox JA (1995). Homicide offending patterns: a grim look ahead. Scientific Proceeding, Annual Meeting of the American Academy for the Advancement of Science. Atlanta. Garbarino J (1995). Raising Children in a Socially Toxic Environment. San Francisco: Jossey-Bass. Greenwood PW: Modell KE, Rvdell CP, Chiesa IR (1996). Diverting Children From a Life of Crime: Measuring Costs and Benefits. Santa Monica, CA: Rand Corporation. Hansen WB. Graham JW (1991). Preventing alcohol, marijuana, and cigarette use among adolescents: peer pressure resistance training versus establishing conservative norms. Prev Med 20:414–430. Hawkins ID, Catalano RF, Morrison DM, O’Donnell J, Abbon RD, Day LE (1992). The Seattle Social Development Project effects of the first four years on protective factors and problem behaviors. In: The Prevention of Antisocial Behavior in Children. McCord J, Tremblay R., eds. New York: Guilford. Henggelet S, Schoenwald S (1994). Boot camps for juvenile offenders just say no. J Child Fam Stud 3:243–248 Institute of Medicine. Committee on Prevention of Mental Disorders (1994). Illustrative prevention intervention research programs. In: Reducing Risks for Mental Disorders: Frontiers for Preventive Intervention Research. Mrazek PJ, Haggerry RJ, eds. Washington, DC: National Academy Press, pp 215–314. Jason J, Flock MT, Tvier CW Jr: (1983). Epidemiology characteristics of primary homicides in the United States. Am J Eptaemtol 117:419–428. Johnson DL (1990). The Houston Parent-Child Development Center Project: disseminating a viable program for enhancing at-risk families. Pres Hum Serv 7:89– 108.
Violent behavior in children and youth
531
Kachur SP, Stennies GM, Powell KE, et al. (1996). School associated violent deaths in the United States, 1992 to 1994. JAMA 275:1729–1733. Kann L, Warren W, Collins JL, Ross J, Collins B, Kolbe LJ (1993). Results from the national school-based 1991 Youth Risk Behavior Survey and progress toward achieving related health objectives for the nation. Public Health Rep 108 (suppl 1):47– 67. Kellam SG, Rebok GW (1992). Building developmental and enological theory through epidemiologically based preventive intervention trials. In: Preventing Antisocial Behavior: Intervention from Birth Through Adolesence. McCord J, Tremblay RE, eds. New York: Guilford, pp 62–195. Lewis DO (1992). From adult to violence: psychological consequences of maltreatment. J Am Acad Child Moles Psychiatry 31:383–391. Lyon JM, Henggeler S, Hall JA (1992). The family relations, peer relations and criminal activities of Caucasian and Hispanic-American gang members. J Abnorm Child Psychol 20:439–449. Mandel MJ Magnussor P, Ellis JE, et al. (1993). The economics of crime. Business Week Dec 13, pp 72–81. Mendel RA (1995). Prevention or Pork?: A Hard-Headed Look at Youth-Oriented AntiCrime Programs., Washington, DC: American Youth Policy Forum. National Association of Childrens Hospitals and Related Institutions (1991). Firearms Fact Sheet. National Center for Health Statistics (1993). Advanced Report of Fatal Mortality Statistics. Hyattsville, MD: National Center for Health Statistics, 42(suppl):20–21. Offord DR, Bennert Kl (1994). Conduct disorder: long-term outcomes and intervention effectiveness. J Am Acad Child Adolesc Psychiatry 33:1069–1078. Olds DL, Henderson CR, Tatelbaum R, Chamberhn R (1988). Improving the litecourse development of socially disadvantaged mothers a randomized trial of nurse home visitation. Am J Public Health 78:1436–1444. Olweu D (1991). Bully victim problems among school children basis facts and effects of an intervention program. In: The Development and Treatment of Childhood Aggression. Rubin K, Pepler D, eds. Hillsdale, NJ: Erlbaum. Ordog G, Wasserberger J Jbanez J, et al (1993). Incidence of gunshot wound at a county hospital following the Los Angeles not and a gang trust Trauma 6:779–781. Osofsky JD (1995). Children who witness domestic violence: the invisible victims. Social Policy Report 9:3. Ann Arbor, MI: Society for Research in Child Development. Pacific Center for Violence Prevention (1993). Preventention Fact Sheet. San Francisco: California Wellness Foundation. Patterson G, Narrett C (1990). The development of a reliable and valid treatment program for aggressive young children. Int J Ment Health 19:19–26. Pitts T, Steiner H (1994). Risk factors for violence and delinquent behavior in a suburban high school. Presented at the Annual Meeting of the American Academy of Child and Adolescent Psychiatry, New York. Rae-Grant N, Thomas F, Offord D, Boyle M (1989). Risk, protective factors and the prevalence of behavioral and emotional disorder in children and adolescents. J Am Acad Child Adoles Psychiatry 28:262–268.
Annual progress in child psychiatry and child development 2000–2001
532
Richters JE, Martinez P (1993) The NIMH community violence report, I: children as victims of and witnesses to violence. Psychiatry 56:7–21. Sampsor R (1998). Urban black violence: the effects of malt robiessness and family disruption. Am J Social 93:348–382. Sloar: JH, Kellermana AL Reay DT, et al (1988). Handgun regulations. Crime, assaults, and homicide: a tale of two cities. N Engl J Med 319:1256–1262. Snyder HW, Sickmund M (1995). Juvenile Offenders and Vicitms: A National Report. Washington, DC: Office of Juvenile Justice and Delinquency Prevention. US Department of Justice (1996). Your Violence: A Community-Based Response. One City’s Success Story. Washington, DC: US Department of Justice. Weikarc DP, Schweinhart LJ, Larner MB (1986). A report on high/scope preschool curriculum comparison study: consequences of three preschool curriculum models through age 15. Early Child Res Q 1:15–45. Weil DS, Knox RC (1996). Effects of limiting handgun purchases on interstate transfer of firearms. JAMA 22:1759–1761. Zigler E (1993). Reshaping early childhood intervention to be a more effective weapon against poverty. In: Proceedings of the 101st Annual Meeting of the American Psychological Association, Toronto, Canada.
PART VI: SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
29 Agents of Change: Pathways through which Mentoring Relationships Influence Adolescents’ Academic Adjustment Jean E.Rhodes, Jean B.Grossman, and Nancy L.Resch
A conceptual model was tested in which the effects of mentoring relationships on adolescents’ academic outcomes were hypothesized to be mediated partially through improvements in parental relationships. The parameters of the model were compared with those of an alternative, in which improved parental relationships were treated as an outcome variable rather than a mediator. The study included 959 young adolescents (M age=12.25 years), all of whom applied to Big Brothers Big Sisters programs. The adolescents were randomly assigned to either the treatment or control group and administered questions at baseline and 18 months later. The hypothesized model provided a significantly better explanation of the data than the alternative. In addition to improvements in parental relationships, mentoring led to reductions in unexcused absences and improvements in perceived scholastic competence. Direct effects of mentoring on global self-worth, school value, and grades were not detected but were instead mediated through improved parental relationships and scholastic competence. Implications of the findings for theory and research are discussed.
INTRODUCTION Volunteer mentoring programs have been advocated increasingly as a means of promoting the academic achievement of adolescents who may be at risk for school failure (Campbell-Whatley, Algozzine, & Obiakor, 1997; Dondero, 1997; Levine & Nidiffer, 1996; Reglin, 1998; Rogers & Taylor, 1997). Indeed, approximately 5 million youth are involved in school- and community-based volunteer mentoring programs nationwide (McLearn, Colasanto, Schoen, & Shapiro, 1999), including more than 100,000 participants in Big Brothers Big Sisters of America programs (McKenna, 1998). Despite the growing popularity of this approach, very little is known about the underlying processes by which mentor relationships affect academic outcomes. In this study, a
Agents of change
535
conceptual model of mentoring was proposed and tested.
BACKGROUND Evaluations of volunteer mentoring programs provide evidence of positive influences on adolescent developmental outcomes, including improvements in academic achievement (McPartland & Nettles, 1991), self-concept, lower recidivism rates among juvenile delinquents (Davidson & Redner, 1988), and reductions in substance abuse (LoSciuto, Rajala, Townsend, & Taylor, 1996). A national evaluation of Big Brothers Big Sisters programs found that in addition to positive changes in grades, perceived scholastic competence, truancy rates, and substance use, mentored youth were more likely than nonmentored youth to report improved parent and peer relationships (Grossman & Tierney, 1998). Although these findings are promising, basic questions remain regarding the underlying factors that may mediate mentors’ influence over time. One possibility is that mentors indirectly affect outcomes through their positive influence on the more proximal relationships in adolescents’ lives. In particular, mentors may bolster the protective effects of parental relationships, which are often strained among youth who are referred to relationship-based interventions (Freedman, 1995; Styles & Morrow, 1995; Tierney, Grossman, & Resch, 1995). Because the central component of mentoring programs is the formation of close alliances between adults and adolescents, mentor relationships can offer a model to adolescents of care and support. In doing so, mentors may challenge negative views that adolescents hold of themselves or of relationships with adults and demonstrate that positive, caring relationships with adults are possible. The helping relationship can thus become a “corrective experience” for those adolescents who may have experienced unsatisfactory relationships with their parents (Olds, Kitzman, Cole, & Robinson, 1997). This experience can then generalize, thereby enabling adolescents to perceive their proximal relationships as more forthcoming and helpful (Coble, Gantt, & Mallinckrodt, 1996; Fairbairn, 1952; Main, Kaplan, & Cassidy, 1985). Support for the potential of positive relationships to modify adolescents’ perceptions of other relationships is derived largely from attachment theory (Bowlby, 1982). According to attachment theorists, children construct cognitive representations of relationships through their early experiences with primary caregivers (Bretherton, 1985). These experience-based expectations, or working models, are believed to be incorporated into the personality structure and to influence behavior in interpersonal relationships throughout and beyond childhood (Ainsworth, 1989; Bowlby, 1988). Although considered to be relatively stable over time, working models are flexible to modification in response to changing life circumstances, such as engagement in unconditionally supportive relationships (Belsky & Cassidy, 1994; Sroufe, 1995). Indeed, with the increases in perspective-taking and interpersonal understanding that often accompany this stage of development, adolescence may lend itself uniquely to the revision of working models (Selman, 1980). As Main et al. (1985, p. 11) have argued, “By adolescence, they [working models] have become quite firm, although new models of thinking here may also provide new opportunities for change.”
Annual progress in child psychiatry and child development 2000–2001
536
Analyses of therapeutic alliances (Bowlby, 1978; Goldfried, 1995; Kohut, 1987), home visitors (Olds et al., 1997), and mentoring relationships (Flaxman, 1997) provide additional support for this process. For example, after intensively examining mentoring relationships, Styles and Morrow (1995) concluded that it was the experience of a trusting and consistently supportive mentor relationship, as opposed to a mentor’s focus on specific goals, that predicted better outcomes among youth. They provided numerous examples of adolescents who developed emotional bonds with their mentors and then gradually began to experience more positive, trusting interactions with their parents and peers. Along similar lines, researchers have found that, in contrast to adolescents who do not have mentors, adolescents with mentors tend to report more satisfying relationships with their parents and other close providers (Hamilton & Darling, 1996; Rhodes, Contreras, & Mangelsdorf, 1994). These positive changes in conceptions of relationships may also facilitate adolescents’ capacity to use mentors as role models and to derive other cognitive and emotional benefits. By conveying messages regarding the value of school and serving as tangible models of success, mentors may stimulate adolescents’ improved attitudes toward school achievement, perceived academic competence, and school performance (Bowman & Howard, 1985; Hamilton & Hamilton, 1990), as well as adolescents’ beliefs about the relationship between educational attainment and future occupational opportunities (Klaw & Rhodes, 1995; Mickelson, 1990). To the extent that adolescents begin to place greater value on school as an important context for attaining future goals, they are expected to achieve academically and behaviorally in that context (Eccles, 1983; Roeser, Midgley, & Urdan, 1996). In addition, through their provision of emotional support and positive feedback, mentors are thought to enhance adolescents’ self-concept (Felson, 1993; Ryan, Stiller, & Lynch, 1994), which, in turn, is related to more positive perceptions of scholastic competence (Covington, 1992; Harter, 1993) and to school-related achievement and behavioral outcomes (Eccles, 1983). Even among youth who generally perceive parental support as available, mentor relationships can alleviate some of the relationship tensions and conflicts that arise throughout adolescence. Alternative sources of adult support can mediate adolescents’ paradoxical needs for both autonomy and adult guidance (Cooper, Grotevant, & Condon, 1983; Hill & Holmbeck, 1986). In addi-tion, by helping the adolescent to cope with everyday stressors, providing a model for effective conflict resolution, and indirectly reducing parental stress, mentor relationships can facilitate improvements in parent-child interactions (Minuchin, 1992; Youniss & Smollar, 1985). Improvements in parental relationships, in turn, can promote improvements in a wide array of outcomes, including the adolescents’ self-worth (Garber, Robinson, & Valentiner, 1998), scholastic competence (Craik, 1997; Klebanov & Brooks-Gunn, 1992; Teachman, Paasch, & Carver, 1996), prosocial behavior (Catalano & Hawkins, 1996; Resnick, Bearman, Blum, et al., 1997), and academic outcomes (Eccles, Early, Fraser, Belansky, & McCarthy, 1997). Lau and Leung (1992), for example, found that better parental relationships were associated with higher levels of academic achievement and self-esteem and lower levels of delinquent behavior. In summary, adolescents’ capacity to benefit from the support of parents and other providers is presumed to be facilitated by the sense of support and acceptance that is
Agents of change
537
derived from mentor relationships. Mentor relationships are expected to improve adolescents’ more proximal, parental relationships, which in turn, should positively influence adolescents’ global self-worth, perceived scholastic competence, school value, grades, and attendance. Additionally, through role modeling and the provision of emotional support and positive reinforcement, mentoring is expected to influence adolescents’ perceptions of self-worth and their beliefs about their competence as learners and their valuing of school. The hypothesized predictive model (Model 1) is illustrated in Figure 29.1. Although there is some support for the validity of the pathways proposed in this model, empirical tests have been largely limited to small-scale, cross-sectional studies. This study makes use of longitudinal data from a large sample of urban adolescents that were collected as part of the national evaluation of Big Brothers Big Sisters, the largest and arguably most influential evaluation of mentoring (Grossman & Tierney, 1998).
Figure 29.1. Model 1.
Annual progress in child psychiatry and child development 2000–2001
538
METHOD Participants The study included 1,138 youth, all of whom applied to Big Brothers Big Sisters programs in 1992 and 1993. Most agencies give preference to youth who have no more than one parent actively engaged in their lives. Other criteria include age (5 through 18), residence in the catchment area, and an agreement by the parent and child to follow agency rules. Applicants were randomly assigned to either the treatment or control group and administered questions at baseline and 18 months later. Eighty-five percent of the sample (N=959, 487 treatments and 472 controls) completed both the baseline and the follow-up interviews. Over half of this analysis sample were boys (62.4%) and approximately half were members of minority groups (57.5%). Seventy-one percent of the minority youth were African Americans, 18% Hispanic, and the remaining were members of a variety of other racial/ethnic groups. Participants ranged in age from 10 to 16 (M=12.25), most (69%) of whom were between the ages of 11 and 13. Ninety percent of youth lived with one parent (94% mothers, 6% fathers), 5% lived with a grandparent, and the remaining participants lived in extended family or nonfamily arrangements. More than 40% of the youth lived in households that were receiving either food stamps or public assistance or both. The only systematic difference between the treatment and control group youth at baseline was that the treatment youth had the opportunity to be matched with a mentor. Design and Procedure From the network of more than 500 Big Brothers Big Sisters local agencies, eight agencies were selected to participate in the outcome study. The key selection criteria for inclusion in the impact study were a large, active caseload, a waiting list, and geographic diversity. With only a few exceptions, all of the youth who enrolled in the eight selected Big Brothers Big Sisters agencies during the intake period were encouraged to participate in the research. Once a youth was informed about the study, determined to be eligible, and assented to participate (along with a parent’s signed, informed consent), he or she was randomly assigned to either the treatment or control group. Only 2.7% of the youth refused to participate in the evaluation. The control group was placed on a waiting list for a poststudy match. All participants were interviewed by telephone before they knew their experimental status. Follow-up interviews were conducted 18 months later by telephone. Agency staff matched particular adult volunteers with particular youth on the basis of gender (only same-sex dyads) and a variety of factors, including shared interest, reasonable geographic proximity, and same-race match preference. All volunteers underwent an intensive screening process, followed by an agencybased training and ongoing case management. The training covered agency policies, communication, and relationship building, as well as issues of particular relevance to participating youth (e.g., grieving, sexual abuse). Dyads typically engaged in a wide variety of leisure- and career-
Agents of change
539
oriented discussions and activities with the general goal of promoting the youth’s positive development. At the conclusion of the study, 378 (78%) of the treatment youth had been matched. Agency staff reported three major reasons for the failure to match the 109 treatment youth during the study period. Thirty-three of the unmatched treatment youth became ineligible during the study period because the parent remarried, the youth was no longer within the eligible age range, or the youth’s place of residence changed. Thirty-one were not matched because the youth no longer wanted a Big Brother or Big Sister. Twenty-one were not matched because a suitable volunteer could not be found during the study period. The 24 remaining treatment youth were not matched for a variety of reasons, most commonly because the parent or youth did not follow through with the intake process. Sixty percent of the matches were still active, whereas 40% were no longer meeting. The ongoing matches had been meeting for an average of 12.9 months, whereas the closed met for an average of 9 months. Over 70% of the youth met with their mentor at least three times a month and approximately 45% met one or more times per week. An average meeting lasted 3.6 hours. Measures Parent Relationships. The Inventory of Parent and Peer Attachment (IPPA; Armsden & Greenberg, 1987) is a 23-item scale containing questions related to a child or adolescent’s relationship with his/her primary caregiver (the corresponding peer questions were not administered). Responses are coded on a 4-point scale, ranging from “hardly ever true” (1) to “very often true” (4). The IPPA contains three subscales: communication (e.g., my mother can tell when I am upset about something), trust (e.g., my father respects my feelings), and alienation (e.g., talking over problems with my mother makes me feel ashamed or foolish). At pretest, Cronbach’s α reliability coefficients of the subscales were .77, .83, and .76, respectively. Only the pretest as are reported. Posttest αs equaled or exceeded pretest as in all instances. Scholastic Competence. This six-item subscale of the Self-Perception Profile for Children (Harter, 1986) contains statements describing confidence in school work that divide children into two groups, for example, “some kids feel that they are very good at their schoolwork/other kids worry about whether they can do the schoolwork assigned to them.” Respondents were asked to determine if they were more like the first or second group and whether the statement was “really true” or “sort of true” for them. Scores ranged from 1 to 4, with higher scores reflecting more positive self-evaluations, α=.77. Grades and Attendance. Individual items relating to scholastic behaviors were asked, including number of unexcused absences from school, grades, visits to college campuses, books read, trips to the library, hours spent on homework, and hours spent reading. For purposes of this study, we focused on the number of unexcused absences and grades, ranging from “mostly D’s and F’s” (1)to “onlyA’s” (8). School Value. This 18-item measure (Berndt & Miller, 1986) assesses the extent to which respondents value academic success and the information that they learn in school, for example, “do you care about doing your best at school?” On a 4-point scale, ranging from “hardly ever” (1) to “pretty often” (4), respondents were asked to indicate the
Annual progress in child psychiatry and child development 2000–2001
540
frequency with which they felt certain ways about school, α=.86. Self-Worth. This six-item subscale of the Self-Perception Profile for Children (Harter, 1986) contains statements describing the global self-worth of two groups, for example, “some kids are pretty pleased with themselves/other kids are often unhappy with themselves.” Respondents were asked to determine whether they were more like the first or second group and whether the statement was “really true” or “sort of true” for them. Scores ranged from 1 to 4, with higher scores reflecting more positive self-evaluations, α=.75. Treatment Status. A youth’s exposure to mentoring was captured by their treatment status, which was coded dichotomously with 0=control and 1= treatment. To avoid biases in the measurement of mentoring effects, both matched and unmatched treatment participants were included in the analyses. If the unmatched treatments were systematically different from their matched counterparts, their exclusion would have biased the impact estimate because the similar control youth were not excluded from the analyses. The only way to obtain an unbiased estimate of mentoring’s impact was to compare the entire treatment group (matched and unmatched) with the entire control group (Rossi & Freeman, 1993).
RESULTS Table 29.1 presents the mean and t-test of the change score difference for each of the six outcomes. Although the two groups were equivalent at baseline, the treatment group youth reported relatively better parental relationships, scholastic competence, and school attendance at follow-up. The other three changes, although not statistically significant, were in the predicted direction. We hypothesized that mentoring would directly impact the adolescents’ perceptions (i.e., global self worth, perceived scholastic competence, value of school) and indices of academic performance and behavior (i.e., grades and unexcused absences), as well as the quality of adolescents’ parental relationships. Parental relationships, in turn, were hypothesized to affect all of the mediators and outcomes, including the adolescents’ perceptions (i.e., global self-worth, perceived scholastic competence, value of school), and academic performance and behavior. Grades were hypothesized to depend on the value that adolescents placed on school, their perceived scholastic competence, and school attendance. Unexcused absences were expected to depend on the value that youth placed on school and their perceived scholastic competence. Because we were interested in explaining changes during the 18-month period and not the level of the outcomes, baseline levels of outcome variables were controlled for in the equation. To test the hypothesis that the impact of mentoring is mediated by its effect on the parental relationship, we compared Model 1 to an alternative, nested model (Model 2) in which the quality of the parental relationship was treated as an outcome variable, like grades and school attendance, rather than a mediator. Perceived scholastic competence, school value, and self-worth were treated as mediators for academic outcomes but not for the quality of the parental relationship (see Figures 29.1 and 29.2). Comparing the relative fit of these two alternative models to the data tests whether the quality of the
Agents of change
541
parental relationship is a mediator through which mentoring affects academic outcomes.
TABLE 29.1 Means and t-Tests of Change Score Differences
Treatment Mean Control Mean t-Test of Difference IPPA .36 (11.8) −1.18 (12.5) −1.96a Global self-worth .96 (4.4) .65 (5.0) 1.01 Value of school −.45 (7.8) −1.41 (8.2) −1.87c Scholastic competence 1.17 (4.6) .22 (4.6) −3.18b Days of school skipped .04 (2.2) .61 (2.9) 3.45b Grades (1=low, 8=high) −.10 (1.8) −.25 (1.8) −1.26 Note. Standard errors in parentheses. ap<.05 bp<.01 cp<.10.
Figure 29.2. Model 2. Both Models 1 and 2 were analyzed by using Lisrel 8 (Jöreskog & Sörbom, 1993). A modification of the models was indicated by the data and was theoretically justifiable, namely, the addition of a path from scholastic competence to the value of school. The addition of this path strengthened the fit of the models but reduced the direct effect of
Annual progress in child psychiatry and child development 2000–2001
542
mentoring on the value of school. To gauge the goodness-of-fit of the two models, we examined several statistics. Although a low, nonsignificant goodness-of-fit c2 statistic corresponds to a better fitting model, it is very sensitive to the sample size and increases as the sample increases. Because our sample is relatively large (N=959) we also examined other goodness-of-fit measures that were designed to gauge how good an approximation the models are in large samples. These other measures are the Root Mean Square Error of Approximation (RMSEA, Steiger, 1990), which measures the discrepancy per degree of freedom; Goodness of Fit Indices (GFI and Adjusted GFI); and Comparative Fit Index (CFI). According to these goodness-of-fit measures, the model fit reasonably well. The χ2 statistic for Model 1 was significant but relatively low, χ2(33, N=959)=77.2, p<.01, and the RMSEA was .038, less than the target .05 level. The GFI was .99, the AGFI was .97, and the CFI was .98, all greater than the target .90. The goodness-of-fit statistics for Model 2 were as follows: χ2(38, N=959)=258.6, p< .001, RMSEA=.08; GFI=.96; AGFI=.91; and CFI=.91. To determine if the parental relationship mediated any part of mentoring impact, we tested whether all the hypothesized paths from IPPA to the other outcomes were jointly equal to 0. This hypothesis was strongly rejected, χ2 (5, N=959)=181, p<.001, which indicates that Model 1 is a significantly better explanation of the data than the alternative Model 2. For this reason, most of the subsequent discussion will center on Model 1. Tables 29.2 and 29.3 present the maximum likelihood parameter estimates of the models. Because the purpose of the paper was to determine whether the effects of mentoring were mediated through improved parental relationships (rather than to compare the relative effects of variables on others), standardized coefficients were not calculated. As indicated on Table 29.2, the hypothesis of a direct influence of mentoring on the youth’s global self-worth, school value, and grades was rejected. On the other hand, the direct effect of mentoring on the youth’s parental relationship, perceived scholastic competence, and skipping school was supported. The quality of the youth’s parental relationship directly affected most of the outcome variables but did not directly affect grades, once the value of school and perceived scholastic competence (both affected by the parent) were controlled. As hypothesized, skipping school had a negative effect on grades.
TABLE 29.2 Maximum Likelihood Estimates of Model 1
Independent Variables Treatment assignment (0=control, 1=treatment) IPPA
IPPA Global SelfWorth .12 .22c
.25c
Dependent Variables Value of Scholastic School Competence .04
.26b
Day of School Skipped −.28c
.26c
.08c
−.09c
Grades
.00
.08
Agents of change
543
Global self-worth 29c Value of school .19c −.11c .02 Perceived .25c .22b scholastic competence Days of school −.18c skipped Note. In addition to the variables included in the table, baseline values of the dependent variables were also included in the relevant equation. ap<.05 bp<.01 cp<.001. Table 29.4 presents the estimated direct and indirect effects of mentoring on the six outcomes, as predicted by Model 1. When both the direct and the indirect effects were accounted for, mentoring led to statistically significant improvements in five of the six outcomes (see Figure 29.3). It led to improvements in the youth’s relationships with their parents, school value, scholastic competence, grades, and reductions in school
TABLE 29.3 Maximum Likelihood Estimates of Model 2
Independent Variables
IPPA Global SelfWorth .17 .22b
Dependent Variables Value of Scholastic School Competence
Day of School Skipped −.14c
Grades
Treatment .10 .08 .27c assignment Global self-worth .32c Value of school .19c −.13c c .00 Perceived .25c .28 scholastic competence Days of school −.18b skipped Note. In addition to the variables included in the table, baseline values of the dependent variables were also included in the relevant equation to control for Time 1 scores. ap<.05 bp<.01 cp<.001.
Annual progress in child psychiatry and child development 2000–2001
544
nonattendance. In the case of the parental relationship, scholastic competence, and school attendance, the positive total effects derived primarily from the direct impact of the mentoring relationship. In the case of the youth’s value of school and their grades, the effect was primarily indirect, through mentoring’s impact on the parental relationship and on the youth’s perceived scholastic competence. The indirect effect of mentoring on global self-worth was statistically significant but small, and thus when it is added to the statistically imprecise direct-effect estimate, we cannot reject the hypothesis that the total effect is zero.
TABLE 29.4 Total, Direct, and Indirect Effects of Mentoring
Outcome IPPA Global self-worth Value of school Scholastic competence Days of school skipped Grades (1=low, 8=high) Note. n.a.=not applicable. ap<.05 bp<.01 cp<.001.
Total Effect 1.51b .39 .78a .79c −.49c .22b
Direct Effect 1.51 .27 .23 .62c −.43c .07
Indirect Effect n.a. .12b .55c .17a −.05a .15c
DISCUSSION The results of this study highlight the benefits of mentoring interventions and validate the hypothesis that improved perceptions of parental relationships, although not the sole determinant, are important mediators of change in adolescents’ academic outcomes and behaviors. Mentoring led to improvements in five of the six hypothesized mediator and outcome variables. It directly affected scholastic competence and school attendance, which suggests that, through role modeling, tutoring, and encouragement, mentors can influence both the cognitive and behavioral dimensions of adolescents’ approach to school. A comparison of the two models suggests that the effects of mentoring are mediated partially through improvements in adolescents’ perceptions of their parental relationships. Whether this occurs through changes in the attachment processes remains undetermined because it is unclear whether the affectional bond that arises within the mentor relationship actually leads to changes in the adolescents’ working model of relationships or simply improves the parental relationship through a reduction in normative developmental tensions. Whatever the underlying processes, it appears that guidance and support from an adult outside of the home can lead to improvements in the quality of the
Agents of change
545
Figure 29.3. Maximum likelihood estimates of Model 1. parent-child relationship. Although most research on parent socialization has focused largely on the characteristics of the parent or child that increase or decrease the quality of the parent-child relationship, this research underscores the importance of examining factors outside this dyad that may be influential. As predicted, improvements in perceptions of parental relationships led to improvements in the value that adolescents placed on school. It is possible that, as the parental relationship improved, the adolescent developed more prosocial values. As such, adolescents may have become more compliant with their parents’ suggestions regarding homework, studying, and attendance. Also as predicted, this shift in values led to less truancy and improved grades. Consistent with previous research, improvements in adolescents’ global self worth were associated with improved perceptions of scholastic competence (Harter, 1993). Mentoring did not directly affect global self-worth but was mediated instead through improved perceptions of parental relationships. It may be the case that mentors’ influence on self-appraisals is more domain specific (i.e., academics) and not captured through general indices of self-worth (DuBois, Felner, Brand, & George, 1999). The findings presented in this study are consistent with this perspective in that there was a direct effect on perceived scholastic competence. It is also possible that because adolescence constitutes a period of identity formation and change, global selfworth may be influenced over a relatively longer period of time (Demo & SavinWilliams, 1992). This would imply that the impact of mentoring on changes in selfconcept might depend on the duration of the relationship. Indeed, mentoring relationships that last 12 months or longer have been found to be associated with significant improvements in adolescents’ self-worth, whereas those with earlier terminations tend to have mild or even negative effects on these domains (Grossman & Rhodes, in press). As
Annual progress in child psychiatry and child development 2000–2001
546
such, future models of mentoring processes should incorporate measures of relationship duration. It will also be important to understand how variations in mentor styles and characteristics affect these pathways and whether the pathways of the model change as a function of such variables as an adolescent’s previous experiences, presenting problems, gender, race, ethnicity, or age. Future models should also incorporate additional mediating variables (e.g., peer relationships) and dependent variables (e.g., psychological outcomes). Finally, it will be important to incorporate multidimensional indices of selfconcept that capture adolescents’ self-appraisals across both domains and relational contexts (Bracken, 1996; DuBois et al., 1999; Harter, Waters, Whitesell, & Kastelic, 1998). The strengths and limitations of the research deserve comment. On the positive side, the use of longitudinal data and structural equation modeling afforded a sensitive test of the hypotheses. Similarly, the large, national sample of adolescents confers confidence in the precision of the parameter estimates and the generalizability of the findings. Nonetheless, the mentor relationships were all situated within the context of a single youth mentoring program. The pattern of findings may thus not apply as well to mentoring interventions that provide volunteers with less training and supervision than is typical of Big Brothers Big Sisters. It is also worth noting that the assessments were based solely on the adolescents’ perceptions. The participants may have been limited in their ability to engage in assessments of their parental relationships or inhibited in their willingness to report personal problems or relationship difficulties. Similarly, participants might have exaggerated their academic progress. It is important to note, however, that self-reported grades have been shown to be an accurate gauge of students’ actual school performance (Johnson, 1975; Sawyer, Laing, & Houston, 1989). Nonetheless, future studies should include additional sources of data. These findings also have implications for the refinement of mentoring interventions. It appears that mentors can positively influence adolescents’ behaviors, school attendance, and sense of competence in school, so the expansion of high-quality mentoring should continue. In light of the crucial role of positive relationships in catalyzing change and the particular vulnerabilities of atrisk adolescents to disappointment and rejection in interpersonal relationships (Downey & Feldman, 1996), such expansion should occur with caution and have sufficient resources to ensure reasonable levels of screening, training, and postmatch mentor support (Sipe, 1996). Additionally, program personnel should remain sensitive to the potential role that parental relationships can play in mediating mentors’ effects and develop ways to capitalize on this function. If parents feel involved in, as opposed to supplanted by, the provision of additional adult support in their children’s lives, they are likely to reinforce mentors’ positive influences. Bowlby (1979, p. 103) has remarked that humans seem “happiest and able to deploy their talents to best advantage when they are confident that, standing behind them, there are one or more trusted persons who will come to their aid should difficulties arise.” To the extent that mentors and parents can work together to provide this backdrop, adolescents are likely to show improvements in multiple domains.
Agents of change
547
ACKNOWLEDGMENTS This study was completed with the assistance of a grant from the William T. Grant Foundation. The authors also gratefully acknowledge the assistance of Joseph Tierney, Jacquelynne Eccles, Regina Langhout, Eva Pomerantz, and Ranjini Reddy and the cooperation of Big Brothers Big Sisters of America.
ADDRESSES AND AFFILIATIONS Corresponding author: Jean E.Rhodes, Harvard University Graduate School of Education, Appian Way, Cambridge, MA 02138; e-mail:
[email protected]. Jean B.Grossman and Nancy L.Resch are with Public/Private Ventures, Philadelphia, PA.
REFERENCES Ainsworth, M. (1989). Attachments beyond infancy. American Psychologist, 44, 709– 716. Armsden, G.C., & Greenberg, M.T. (1987). The Inventory of Parent and Peer Attachment (IPPA): Relationship to well being in adolescence. Journal of Youth and Adolescence, 16, 427–454. Belsky, J., & Cassidy, J. (1994). Attachment and close relationships: An individualdifference perspective. Psychological Inquiry, 5, 27–30. Berndt, T., & Miller, K. (1986). Expectancies, values, and achievement in junior high school. Journal of Educational Psychology, 82, 319–326. Bowlby, J. (1978). Attachment theory and its therapeuticimplications. Adolescent Psychiatry, 6, 5–3. Bowlby, J. (1979). The making and breaking of affectional bonds. London: Tavistock. Bowlby, J. (1982). Attachment and loss: Retrospect and prospect. American Journal of Orthopsychiatry, 52, 664–676. Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human development. New York: Basic Books. Bowman, P.J., & Howard, C. (1985). Race-related socialization, motivation, and academic achievement: A study of Black youths in three-generation families. Journal of the American Academy of Child Psychiatry, 24, 134–141. Bracken, B.A. (1996). Clinical applications of a contextdependent, multidimensional model of self-concept. In B.A.Bracken (Ed.), Handbook of self-concept: Developmental, social, and clinical considerations (pp. 463–503). New York: Wiley. Bretherton, I. (1985). Attachment theory: Retrospect and prospect. In I.Bretherton & E.Waters (Eds.), Growing points of attachment theory and research (pp. 3–35). Monographs of the Society for Research in Child Development, 50(1–2, Serial No. 209).
Annual progress in child psychiatry and child development 2000–2001
548
Campbell-Whatley, G.D., Algozzine, B., & Obiakor, F. (1997). Using mentoring to improve academic programming for African American male youths with mild disabilities. School Counselor, 44, 362–367. Catalano, R.F., & Hawkins, J.D. (1996). The social development model: A theory of antisocial behavior. In J.David Hawkins (Ed.), Delinquency and crime: Current theories. Cambridge criminology series (pp. 149–197). New York: Cambridge University Press. Coble, H.M., Gantt, D.L., & Mallinckrodt, B. (1996). Attachment, social competency, and the capacity to use social support. In G.R.Pierce, B.R.Sarason, & I. G.Sarason (Eds.), Handbook of social support and the family. New York: Plenum Press. Cooper, C.R., Grotevant, H.D., & Condon, S.M. (1983). Individuality and connectedness both foster adolescent identity formation and role taking skills. In H.D. Grotevant & C.R., Cooper (Eds.), Adolescent development in the family: New directions for child development (pp. 43–59). San Francisco: Jossey-Bass. Covington, M.V. (1992). Making the grade: A self-worth perspective on motivation and school reform. New York: Cambridge University Press. Craik, L.R. (1997, November). Attachment to parents and peers in early adolescence: Relationships to perceived schoolastic competence and academic achievement. Dissertation Abstracts International Section A: Humanities & Social Sciences, 58, 1577. Davidson, W.S., & Redner, R. (1988). The prevention of juvenile delinquency: Diversion from the juvenile justice system. In R.H.Price, E.L.Cowen, R.P.Lorion, & J.RamosMcKay (Eds.), Fourteen ounces of prevention: Theory, research, and prevention (pp. 123–137). New York: Pergamon. Demo, D.H., & Savin-Williams, R.C., (1992). Self-concept stability and change during adolescence. In R.P.Lipka & T.M.Brinthaupt (Eds.). Self-perspectives across the life span (pp. 116–150). Albany: State University of New York Press. Dondero, G.M. (1997). Mentors: Beacons of hope. Adolescence, 32, 881–886. Downey, G., & Feldman, S.I. (1996). The implications of rejection sensitivity for intimate relationships. Journal of Personality and Social Psychology, 70, 1327–1343. DuBois, D.L., Felner, R.D., Brand, S., & George, G.R. (1999). Profiles of selfesteem in early adolescence: Identification and investigation of adaptive correlates. American Journal of Community Psychology, 27, 899–932. Eccles, J.S. (1983). Expectancies, values, and academic behaviors. In J.T.Spence (Ed.), The development of achievement, promotion (pp. 283–331). Greenwich, CT: JAI Press. Eccles, J.S., Early, D., Fraser, K., Belansky, E., & McCarthy, K. (1997). The relation of connection, regulation, and support for autonomy to adolescents’ functioning. Journal of Adolescent Research, 12, 263–286. Fairbairn, W.R.D. (1952). An object-relation theory of the personality. New York: Basic Books. Felson, R.B. (1993). The (somewhat) social self: How others affect self-appraisals. In J.Suls (Ed.), Psychological perspectives on the self (pp. 1–26). Hillsdale, NJ: Erlbaum. Flaxman, E. (1997). Youth mentoring: Programs and practices. (ERIC Clearinghouse on Urban Education, No. 97) New York: Columbia University. Freedomn, M. (1995). The kindness of strangers: Adult mentors, urban youth, and the
Agents of change
549
new volunteerism. San Francisco: Jossey-Bass. Garber, J., Robinson, N.S., & Valentiner, D. (1998). The relation between parenting and adolescent depression: Selfworth as a mediator. Journal of Adolescent Research, 12, 12–33. Goldfried, M.R. (1995). From cognitive-behavior therapy to psychotherapy integration: An evolving view. New York: Springer Publishing. Grossman, J.B., & Rhodes, J.E. (in press). The test of time: Length of mentoring relationship as predictor of adolescent outcomes. American Journal of Community Psychology. Grossman, J.B., & Tierney, J.P. (1998). Does mentoring work? An impact study of the Big Brothers/Big Sisters. Evaluation Review, 22, 403–426. Hamilton, S.F., & Darling, N. (1996). Mentors in adolescents lives. In K.Hurrelmann & S.F.Hamilton (Eds.) Social problems and social contexts in adolescence: Perspectives across boundaries (pp. 199–215). New York: Aldine De Gruyter. Hamilton, S.F., & Hamilton, M.A. (1990). Linking up: Final report on a mentoring program for youth. New York: Cornell University, College of Human Ecology. Harter, S. (1986). Cognitive-developmental processes in the integration of concepts about emotions and the self [Special issue: Developmental perspectives on social-cognitive theories]. Social Cognition, 4, 119–151. Harter, S. (1993). Causes and consequences of low self-esteem in children and adolescents. In R.F.Baumeister (Ed.), Self-esteem: The puzzle of low self-regard. New York: Plenum Press. Harter, S., Waters, P.L., Whitesell, N., & Kastelic, D. (1998). Level of voice among female and male high school students: Relational context, support, and gender orientation. Developmental Psychology, 34, 892–901. Hill, J.P., & Holmbeck, G.N. (1986). Attachment and autonomy during adolescence. In G.J.Whitehurst (Ed.), Annals of child development (pp. 145–189). Greenwich, CT: JAI Press. Johnson, P.B. (1975). Achievement motivation and selfreported grade point average. Psychology in the Schools, 12, 402–404. Jöreskog, K.G., & Sörborn, D. (1993). Lisrel 8: Structural equation modeling with the SIMPLIS command language. Hillsdale, NJ: Erlbaum. Klaw, E.L., & Rhodes, J.E. (1995). Mentor relationships and the career development of pregenant and parenting African American teenagers. Psychology of Women Quarterly, 19, 551–562. Klebanov, P., & Brooks-Gunn, J. (1992). Impact of maternal attitudes, girls’ adjustment, and cognitive skills upon academic performance in middle and high school. Journal of Research on Adolescence, 2, 81–102. Kohut, H. (1987). Extending empathic understanding, sharing an attitude. In Elson, M. (Ed.), The Kohut seminars on self psychology and psychotherapy with adolescents and young adults. New York: Norton. Lau, S., & Leung, K. (1992). Relations with pareents and school and Chinese adolescents’ self-concept, delinquency, and academic performance. British Journal of Educational Psychology, 62, 193–202. Levine, A., & Nidiffer, J. (1996). Beating the odds: How the poor get to college. San
Annual progress in child psychiatry and child development 2000–2001
550
Francisco: Jossey-Bass. LoSciuto, L., Rajala, A.K., Townsend, T.N., & Taylor, A.S. (1996). An outcome evaluation of Across Ages: An intergenerational mentoring approach to drug prevention. Journal of Adolescent Research, 11, 116–129. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I.Bretherton & E.Waters (Eds.), Growing points of attachment theory and research (pp. 66–104). Monographs of the Society for Research in Child Development, 50 (1–2, Serial No. 209). McKenna, T. (1998, June). Big Brothers/Big Sisters of America: A view from the national level. Papers presented at The State and Future of Mentoring Symposium. Washington, DC. McLearn, K.T., Colasanto, D., Schoen, C., & Shapiro, M.Y. (1999). Mentoring matters: A national survey of adults mentoring young people. In J.B.Grossman (Ed.), Contemporary issues in mentoring (pp. 66–83). Philadelphia: Public/Private Ventures. McPartland, J.M., & Nettles, S.M. (1991). Using community adults as advocates or mentors for at-risk middle school students: A two-year evaluation of project RAISE. American Journal of Education, 99, 568–586. Mickelson, R.A. (1990). The attitude-achievement paradox among black adolescents. Sociology of Education, 63, 44–61. Minuchin, P. (1992). Conflict and child maltreatment. In C.U.Shantz & W.W.Hartup (Eds.), Conflict in child and adolescent development. Cambridge studies in social and emotional development (pp. 380–401). New York: Cambridge University Press. Olds, D., Kitzman, H., Cole, R., & Robinson, J. (1997). Theoretical formulations of a program of home visitation for pregnant women and parents of young children. Journal of Community Psychology, 25, 9–26. Reglin, G. (1998). Mentoring students at risk: An underutilized alternative education strategy for K-12 teachers. Springfield, IL: Charles C.Thomas. Resnick, M.D., Bearman, P.S., Blum, R.W., Bauman, K.E., Harris, K.M., Jones, J., Tabor., J., Beuhring, T., Sieving, R.E., Shew, M., Ireland, M., Bearinger, L.H., & Udry, J.R. (1997, September 10). Protecting adolescents from harm: Findings from the National Longitudinal Study on Adolescent Health. Journal of the American Medical Association, 278, 823–832. Rhodes, J.E., Contreras, J.M., & Mangelsdorf, S.C. (1995). Natural mentor relationships among Latina adolescent mothers: Psychological adjustment, moderating processes, and the role of early parental acceptance. American Journal of Community Psychology, 22, 211–228. Roeser, R.W., Midgley, C., & Urdan, T.C. (1996). Perceptions of the school psychological environment and early adolescents’ psychological and behavioral functioning in school: The mediating role of goals and belonging. Journal of Educational Psychology, 88, 408–422. Rogers, A.M., & Taylor, A.S. (1997). Intergenerational mentoring: A viable strategy for meeting the needs of vulnerable youth. Journal of Gerontological Social Work, 28, 125–140. Rossi, P.H., & Freeman, H. (1993). Evaluation: A systematic approach. Newbury Park, CA: Sage.
Agents of change
551
Ryan, R.M., Stiller, J.D., & Lynch, J.H. (1994). Representations of relationships to teachers, parents, and friends as predictors of academic motivation and self-esteem. Journal of Early Adolescence, 14, 226–249. Sawyer, R., Laing, J., & Houston, W. (1989). Accuracy of self-reported high school courses and grades of collegebound students. College and University, 26, 280–299. Selman, R. (1980). The growth of interpersonal understanding. New York: Academic Press. Sipe, C.L. (1996). Mentoring: A synthesis of P/PV’s Research: 1988–1995. Philadelphia: Public/Private Ventures. Sroufe, A.L. (1995). Contribution of attachment theory to developmental psychopathology. In E.A.Carlson & A.L.Sroufe (Eds.), Developmental psychopathology: Vol. 1. Theory and methods. New York: Plenum Press. Steiger, J.H. (1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180. Styles, M.B., & Morrow, K.V. (1995). Understanding how youth and elders form relationships: A study of four linking lifetimes programs. Philadelphia: Public/ Private Ventures. Teachman, J.D., Paasch, K., & Carver, K. (1996). Social capital and dropping out of school early. Journal of Marriage and the Family, 58, 773–783. Tierney, J.P., Grossman, J.B., & Resch, N.L. (1995). Making a difference: An impact study of Big Brothers/Big Sisters. Philadelphia: Public/Private Ventures. Youniss, J., & Smollar, J. (1985). Adolescent relations with mothers, fathers, and friends. Chicago: University of Chicago Press.
PART VI: SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION
30 Initial Impact of the Fast Track Prevention Trial for Conduct Problems: II. Classroom Effects Conduct Problems Prevention Research Group
This study examined the effectiveness of the universal component of the Fast Track prevention model: the PATHS (Promoting Alternative THinking Strategies) curriculum and teacher consultation. This randomized clinical trial involved 198 intervention and 180 comparison classrooms from neighborhoods with greater than average crime in four U.S. locations. In the intervention schools, Grade 1 teachers delivered a 57-lesson social competence intervention focused on self-control, emotional awareness, peer relations, and problem solving. Findings indicated significant effects on peer ratings of aggression and hyperactive-disruptive behavior and observer ratings of classroom atmosphere. Quality of implementation predicted variation in assessments of classroom functioning. The results are discussed in terms of both the efficacy of universal, school-based prevention models and the need to examine comprehensive, multiyear programs.
INTRODUCTION As models of prevention have become better integrated with research on the development of antisocial behavior, the need for, addressing the transactional nature of risk factors has emerged. As noted by the Institute of Medicine (1994), “the transactional interaction between the individual child and his or her environment over time is the ecological crucible in which the pathways to the development of a range of positive or negative childhood and adult outcomes are forged” (p. 181). A critical factor in the early development of antisocial behavior for many high-risk children is that they attend schools that have a high density of other high-risk children like themselves and thus present the classroom teacher, who often has fewer resources than do teachers in less high-risk schools, with additional challenges to classroom order. This volatile combination of highrisk child and high-risk classroom has negatively synergistic consequences for both the child and the classroom. The disruptive behavior of high-risk children undermines the
Annual progress in child psychiatry and child development 2000–2001
554
social and academic environment for other children. Reciprocally, the high-risk child encounters greater stimulation to disruptive behavior and a less nurturant learning atmosphere, rather than being given the kind of well-structured and nonprovocative classroom that would be needed to compensate for his or her emotional, social, and cognitive deficits. The primary goal of the Fast Track model is to integrate the provision of universal (all children) and selective (children at some risk) services into a comprehensive model that involves the child, school, family, and community (Institute of Medicine, 1994). Fast Track was designed to provide two levels of child intervention simultaneously during the elementary school years. Through a multistage screening process involving both teacher and parent reports during kindergarten, the 10% of children demonstrating the greatest degree of early conduct problems were selected for a series of interventions that included weekly parenting support classes, small-group social skills interventions, academic tutoring, and home visiting (Bierman, Greenberg, & Conduct Problems Prevention Research Group [CPPRG], 1996; McMahon, Slough, & CPPRG, 1996). Such interventions are believed to be necessary to both reduce risk factors and promote protective factors in children who are showing the “early starter” model of conduct problems (CPPRG, 1992; Loeber, 1990; Moffitt, 1993). Simultaneous with the initiation of these interventions for the high-risk children and families, the universal intervention consisting of the Fast Track PATHS (Promoting Alternative THinking Strategies) curriculum and behavioral consultation was started in the classroom. There are two central reasons that integrated delivery of universal and selective interventions should provide an additive effect. First, it is unlikely that effects of the selective interventions with the children and families will generalize to the school and classroom setting without providing support for these new skills in the school (Kazdin, 1990, 1993). By providing similar skills, cues, and a common language in both the selective and universal interventions, teachers and other school staff are able to promote the generalization of skills to the classroom. Second, a universal intervention intended to promote the development of social competence in all children should lead to an improved classroom atmosphere that supports improved interpersonal relations for all students (Battistich, Solomon, Watson, Solomon, & Schaps, 1989; Elias et al., 1998). Reciprocally, more intensive intervention with the highest risk children in these same classrooms may serve to reduce their highly disruptive impact on the classrooms, thereby making it easier for the remaining children to respond to the universal intervention. It was hoped that intensive intervention for higher need students would serve to communicate to teachers that the program staff were committed to helping them with their most difficult problems, thus making teachers more open to participating in a universal intervention. In addition, Fast Track staff provided behavioral consultation to teachers regarding both the high-risk children and the remaining classroom students. In this report, we consider the effectiveness of the universal intervention using a randomized clinical trial design. We examine as effects at the level of the classroom (Murray & Wolfinger, 1994) in altering the conditions of peer relations and classroom atmosphere through the use of multiple reporters (children, teachers, and observers). Although sets of schools, not classrooms, were the unit of randomization, the classroom is the unit of program delivery and also the level at which measures such as dosage and
Initial impact of the fast track prevention trial for conduct problems
555
implementation can be assessed. Social competence can be broadly conceived as the capacity to integrate cognition, affect, and behavior to achieve specified social tasks and positive developmental outcomes (Waters & Sroufe, 1983; Weissberg & Greenberg, 1998). Meta-analyses of universal evention programs during the elementary school years indicate that such programs have shown significant and moderate effects on social-cognitive abilities and mild to moderate effects on children’s social adjustment (Beelmann, Pfingsten, & Losel, 1994; Abraham & Almeida, 1987; Durlak, 1995; Schneider, 1992). A number of other conclusions have been reached. First, traditional prevention models based on single skills (e.g., only social problem solving, self-control, empathy, or behavioral social skills training) have demonstrated less effectiveness than multimodal programs that integrate social problem solving, social skills (peer relations and self-control), and emotional understanding. Second, interventions need to be of significant duration and intensity to show effects (Weissberg & Elias, 1993). Third, there is a need to examine how both dosage and quality of implementation affect outcomes. The PATHS preventive intervention program is based on the ABCD (Affective-Behavioral-Cognitive-Dynamic) model of development (Greenberg & Kusche, 1993; Greenberg, Kusche, Speltz, 1991), which places primary importance on the developmental integration of affect (and emotion language), behavior, and cognitive understanding as they relate to social and emotional competence. A basic premise is that a child’s coping, as reflected in his or her behavior and internal regulation, is a function of emotional awareness, affective-cognitive control, and social-cognitive understanding. Between the ages of 5 and 7 years, children undergo a major developmental transformation that generally includes increases in cognitive skills, as well as changes in brain size and function (Kendler, 1963; Pennington & Welsh, 1995; White, 1965). This transition and accompanying developmental changes allow children to undertake major changes in responsibilities, independence, and social roles (Belsky & MacKinnon, 1994; Pianta. Steinberg, & Rollins, 1995). Thus, the relationships among affective understanding, cognition, and behavior are of crucial importance in socially competent action and healthy peer relations (Weissberg & Elias, 1993). Taking this into account, the PATHS curriculum model synthesizes the domains of self-control, emotional awareness and un-derstanding, peer-related social skills, and social problem solving to increase social and emotional competence. In addition to a person-oriented model that focuses primarily on developmental integration, the intervention model incorporates an ecobehavioral systems orientation (Weissberg, Caplan, & Sivo, 1989), which places primacy on the manner in which the teacher uses the curriculum model. That is, program impact may be the greatest when teachers generalize support for curriculum-based skills during the day and build a healthy classroom atmosphere that supports the child’s skill use and internalization of skills. It is presumed that improvements in social competence can be a function of changes in the child, changes in the ecology, and their interaction. Previous field trials with different versions of the PATHS curriculum with both deaf (Greenberg & Kusche, 1993, 1998) and regular- and special-needs children (Greenberg & Kusche, 1997; Greenberg, Kusche, Cook, & Quamma, 1995) have shown that the use of the PATHS curriculum is associated with significantly more mature social cognitions, including better understanding of social problems, higher percentages of effective
Annual progress in child psychiatry and child development 2000–2001
556
solutions, lower percentages of aggressive and passive solutions, and increased recognition of emotions. In all three samples studied to date, teachers reported significant improvements in behaviors targeted by PATHS (self-control, emotional understanding, and use of more effective conflict resolution skills). In special-needs children, PATHS also led to significant decreases in self-reported sadness, decreases in teacher reports of internalizing problems, and increases in teacher reports of social competence. Following from the model, the Fast Track universal prevention strategy operated under the following assumptions. First, the school environment is a fundamental ecology and one that can be a central locus of change (Elias et al., 1998). Second, given the developmental issues facing children in middle childhood, self-control, emotional understanding and awareness, peer-related social skills, and problem solving would all appear to be prime targets for preventive intervention during the elementary school years. Following in the tradition of numerous recent school-based preventive interventions (Elias, Gara, Schuyler, Branden-Muller, & Sayette, 1991; Hawkins & Weis, 1985; Weissberg & Elias, 1993), the PATHS curriculum was designed to be delivered by teachers with support from project staff, to be taught on a regular basis throughout most of the school year, and to provide daily activities to promote generalization. Using an ecobehavioral systems model, Fast Track staff also consulted with the school principal to bring the philosophy of PATHS to the entire school, resulting in various efforts (on a school-by-school basis), such as placing PATHS posters in school hallways, implementing new school behavior guidelines, and painting problem-solving “stoplights” on school playgrounds.
TABLE 30.1 Means by Site and by Condition of School-Level Variables Indicating Poverty, Ethnicity, and Achievement
Site and Condition Durhama Intervention Control Nashville Intervention Control Rural PA Intervention Control Seattle Intervention Control
Percent of Children Receiving Free or Reduced Lunch M SD
Percent of Minority Children M SD
Reading Percentile Score M SD
Math Percentile Score M SD
83.8 75.5
12.5 21.2
90.8 89.7
10.4 17.8
139.9 139.0
— —
135.2 135.5
— —
78.5 77.0
12.4 10.9
61.0 47.3
22.2 23.8
30.0 36.4
9.5 9.2
32.0 37.5
10.6 13.2
39.6 39.1
16.4 13.4
1.0 1.0
0.6 0.9
63.1 52.3
16.5 14.0
60.8 56.8
13.6 15.8
45.4 46.6
7.0 14.2
50.1 53.9
17.3 22.3
47.0 44.9
5.7 5.4
49.2 48.7
5.8 5.7
Initial impact of the fast track prevention trial for conduct problems
557
Note. PA=Pennsylvania. aNorth Carolina schools have initiated their own achievement testing system that presently has no percentile scores and thus cannot be compared with the other sites. The present study advances the knowledge base on school-based universal prevention programs in four ways. First, it examines the efficacy of a universal program in the context of a comprehensive model that includes targeted intervention. Second, it uses a model in which emotional understanding and regula-tion are a central focus of intervention. Third, because of its large sample size, it is the first elementary school study in which analyses can be accomplished at the classroom level. This provides two innovations: There is sufficient power to assess measures of the classroom atmosphere, and hierarchical modeling can be used to account for interdependency among scores within classrooms (Bryk & Raudenbush, 1992). Finally, the study can assess the effects of curriculum effectiveness in sites with different populations. We hypothesized that the introduction of the Fast Track universal prevention program would lead to differences in peer-rated aggression and hyperactive-disruptive behavior, teacher-rated aggression and conduct problems, and observer-rated classroom atmosphere. Further, we hypothesized that both dosage and quality of implementation would predict variation in the preceding outcomes.
METHOD Participants The participating schools were selected from four areas of the country, each representing a different cross-section of the U.S. population: (a) Durham, North Carolina, a small city with a large low- to middle-SES (social economic status), primarily African-American school population; (b) Nashville, Tennessee, a moderate-sized city with a mix of low- to middle-SES, African-American and European-American families; (c) Seattle, Washington, a moderate-sized city with a low- to middle-SES, ethnically diverse population; and (d) central Pennsylvania, a mostly rural area with low- to middle-SES, European-American families. In the Seattle site, both an urban and a suburban district were chosen; in rural Pennsylvania, three small school districts participated. Within each site, approximately 12 elementary schools in high-risk neighborhoods (or towns in the case of rural Pennsylvania) were invited to be involved in the Fast Track intervention model.1 High-risk status was defined from estimated rates of delinquency and juvenile arrest in the neighborhoods. The full Fast Track prevention model was initially described to principals and teachers at each school. After faculty discussion, 1
Because Nashville schools were substantially larger in size, fewer schools were picked at this site.
Annual progress in child psychiatry and child development 2000–2001
558
school-based decisions were made regarding participation. Schools were aware that once they decided to participate, they had a 50% chance of being randomized as a comparison school. After obtaining consensus to participate, we divided schools into matched “sets,” which were equivalent on school sizes, achievement levels, poverty, and ethnic/racial diversity. These sets of schools were then randomly assigned to intervention and control groups. The intervention was conducted in 3 successive years with three cohorts of first graders. There were 198 intervention classrooms and 180 matched comparison classrooms across the three cohorts.2 The total number of children for whom we obtained consent in these classroom was 7,560; 845 of these children were either high-risk intervention or highrisk control children.3 Thus, in analyses that do not include the highrisk children, the sample size was 6,715. Although there were substantial differences between sites in the degree of risk shown by their respective school locations, there was considerable risk in the typical school selected for this intervention. Table 30.1 presents information by site and condition for the variables of poverty, ethnicity, and achievement. The percentage of children receiving free or reduced lunch was 55% (ranging from 39% in rural Pennsylvania to 80% in Durham). The mean percentage of ethnic minority children (primarily African-American) attending the schools was 49% (ranging from 1% in rural Pennsylvania to 90% in Durham). The mean reading percentile across three of the sites (excluding Durham)4 was the 47th percentile (ranging from the 33rd percentile in Nashville to the 57th percentile in rural Pennsylvania). A series of analyses of variance indicated no significant differences between intervention and control schools on the percentage of children who received free or reduced lunch, percentage of ethnic minority children, or academic achievement scores. Approximately 25% of all teachers were of an ethnic minority, and this did not vary by intervention status; 98% of all teachers were female. The Intervention The Fast Track PATHS curriculum in Grade 1 contained 57 lessons, approximately 80% of which were drawn from the published version of the curriculum (Kusche & Greenberg, 1994). Previously designed for special-needs populations, this multiyear (Grades 1 to 5) classroom prevention program was adapted to fit the needs of regular-education students in high-risk schools for the Fast Track program. 2
These classrooms were regular-education classrooms in which PATHS was taught or which were designated as control classrooms. In the accompanying report (CPPRG, 1999a), the number of classrooms differs because that report includes those classrooms (regular and special education) in which high-risk children were present. 3 There were 891 intervention and control children in the high-risk sample described in the accompanying article (CPPRG, 1999a). However, high-risk children who were in a specialeducation classroom or who had moved to a different school were not included in the highrisk sample in this article. 4 North Carolina schools have initiated their own achievement testing system that presently has no percentile scores and thus cannot be compared with other sites.
Initial impact of the fast track prevention trial for conduct problems
559
In the Fast Track version of PATHS, about 40% of the lessons focus on skills related to understanding and communicating emotions. As a basic step toward self-control, PATHS teaches young children to recognize the internal and external cues of affect and to label them with appropriate terms. In a series of lessons, “feeling” words are identified, along with descriptions of the sorts of situations that may elicit the feeling, descriptions of the external cues to recognize that feeling in others, and the internal cues to identify that feeling in oneself. Additional lessons help children understand the difference between feelings and behaviors. Appropriate and inappropriate behavioral responses are discussed. The teaching of feelings involves a generalization technique (“Feeling Faces”) used to promote the student’s use of new knowledge and skills throughout the classroom day. After each emotion concept is introduced, the children personalize their own Feeling Face for that affect; these faces are small cards with idealized line drawings of the affect that are kept on the students’ desks. The faces allow the children to communicate their feelings with minimal difficulty throughout the day, and they facilitate the children’s understanding about how feelings change. Teachers have their own set of Feeling Faces and use them as models for their students. Teachers are encouraged to promote generalization at the beginning and at the end of the day, after recesses, and after lunch time by suggesting that the children evaluate how they feel and display the appropriate face(s). Another 30% of the lessons focus on skills related to the increase of positive social behavior (e.g., social participation, prosocial behavior, and communication skills). Lessons address making and sustaining friendships, using good manners, taking turns and sharing in games, expressing one’s viewpoint, and listening to others. In addition, positive behaviors are elicited and reinforced during each lesson. For example, during each lesson, one child serves as the teacher’s helper (the “PATHS Kid of the Day”), and this child receives compliments from classmates, the teacher, and him- or herself. Finally, about 30% of the lessons focus on self-control and other steps in social problem solving. The development of self-control, affective awareness and communication, and beginning problem-solving skills are integrated with the introduction of the Control Signals Poster (CSP). The CSP is designed like a traffic signal and is a modified version of the Stop Light used in the YaleNew Haven Middle-School Social Problem Solving Program (Weissberg, Caplan, & Bennetto, 1988). The CSP has a red light to signal “Stop—Calm Down,” a yellow light for “Go Slow—Think,” a green light to signal “Go—Try My Plan,” and at the bottom the words” Evaluate—How Did My Plan Work?” Children are taught that when they are in a situation that they find upsetting or frustrating (such as a playground conflict or difficult work situations), the first step toward effective problem solving is to “go to the red light” in order to stop and think before they act. Before they take an action, they should “take a long, deep breath,” calm down, and “say the problem and how they feel.” Once the problem is identified, they can move to the yellow light to “Make a Plan,” considering first the possible solutions and then selecting the best option. The next step is to “Try the Plan” at the green light and evaluate the effectiveness of that plan, recycling through the problem-solving steps if the plan proves ineffective. In addition to scripted lessons teaching children these steps to problem solving, teachers are taught how to hold classroom problem-solving meetings to help children use the problem-solving steps to address current classroom problems.
Annual progress in child psychiatry and child development 2000–2001
560
Typically, skill concepts are presented through direct instruction, discussion, modeling stories, or video presentations. Discussion and role-playing activities follow, giving children a chance to practice the skill and teachers a chance to monitor the level of understanding and skill attained by each class. Although a standard script describes each lesson, teachers are encouraged to adjust the level of presentation and amount of practice as dictated by the responsivity and developmental level of each class. Although the lessons form an important part of the PATHS program, teachers are also encouraged to generalize their use of PATHS concepts across the school day and to other settings of the school outside the classroom. In particular, teachers are encouraged to help children identify their feelings, communicate clearly with others, use self-control strategies, and apply the three steps of problem solving as frustrations, challenges, and interpersonal problems occur at school. Each classroom has a mailbox in which students can submit problems or concerns that are then discussed in problem-solving meetings. To generalize concepts to the home situation, the curriculum includes frequent parent updates on curriculum content and suggestions for ways parents can promote their children’s growing competence. Regular homework activities are designed to help children engage their parents in cooperative activities, such as completing drawings or sharing stories related to curriculum components. Teacher Training The intervention teachers attended a 2.5-day training workshop and received weekly consultation and observation from project staff. The PATHS lessons were taught approximately two to three times per week, with each lesson lasting 20 to 30 minutes from mid-September to May. Teachers were either paid for their extra preparation and consulation time or received continuing education credit for their participation. The weekly consultations were intended to enhance the quality of implementation through modeling, coaching, and providing ongoing feedback regarding program delivery. Fast Track staff (termed educational coordinators [ECs]) also provided general feedback on classroom and behavior management. The ECs were experienced teachers hired by the project. They spent an average of 1 to 1.5 hours per week in each classroom observing, demonstrating, or team teaching the PATHS lessons. They also met individually or in groups with teachers on a regular basis. Measures of Intervention Dosage and Quality of Implementation To assess the amount of dosage, we had teachers report weekly to their assigned ECs on the lessons they had presented. The mean number of lessons taught by Grade 1 teachers was 48.2 (SD=9.7, range=13 to 57). Fidelity was assessed through monthly ratings of quality of implementation made by the ECs on the basis of their direct observation of teacher instruction. For all three cohorts, there were four 4-point Likert-scale ratings (ranging from low-skilled to highly skilled performance). The four ratings were: (a) quality of teaching of PATHS concepts; (b) modeling of PATHS concepts throughout the day; (c) quality of classroom management (during PATHS lessons); and (d) openness to consultation from the EC. Data aggregation across the year indicated that these measures
Initial impact of the fast track prevention trial for conduct problems
561
were highly consistent over time (α=.88); thus, a mean score for each rating was computed for each teacher. The first three ratings were moderately correlated (range=.50 to .70), and the fourth rating correlated less strongly with the other three (range=.30 to .35). The four ratings were kept separate for analyses because they are conceptually distinct domains. Ratings were not available for 14 of the intervention classrooms (7%) because either the teacher changed during the year (e.g., medical/pregnancy leaves) or, in six cases, teachers did not allow ECs to observe their classrooms regularly. It was not feasible to conduct adequate interrater reliability because of both the size of the sample (198 intervention classrooms) and the intensity of data collection; these ratings required knowledge of the curriculum and observations both during and outside of the lesson time and during consultation meetings. In a previous, smaller study of PATHS, adequate reliability (r=.63 to .85) of these ratings had been demonstrated (Cheney, Greenberg, & Kusche, 1991). Measures of Outcome Outcome measures were derived from three independent sources. The measures presented here are the only measures available on the entire classroom population. Measures that focused on the targeted population were more extensive and are presented in a companion article (CPPRG, 1999a). First, at the beginning and end of Grade 1, teachers were individually interviewed regarding the behavior of each child in their class using the Teacher Observation of Classroom Adaptation—Revised (TOCA-R; Werthamer-Larsson, Kellam, & Wheeler, 1991) and the Social Health Profile (SHP; CPPRG, 1999b). Second, in the spring of Grade 1, sociometric assessments were collected to assess peer aggression, peer hyperactivity-disruptiveness, and peer social status. Third, also in the spring of Grade 1, observes who were unaware of the status of the schools assessed the quality of the classroom atmosphere using a 10-item scale. Neither the sociometric nor observational measures were used as a pretest assessment. Both measures would have required a month of school adaptation and 2 months to collect; this would have delayed the program, which was designed to begin as early in the school year as possible. Teacher Report. Grade 1 teachers completed the TOCA-R and the SHP in the fall and spring during a structured interview. On the TOCA-R, teachers rated the behavior of each child in the class on items using 6-point Likert scales ranging from 0 (almost never) to 5 (almost always). The interview covering the entire class required about 90 min to complete, and teachers were reimbursed for their time. Two internally consistent factors from the TOCA-R were used in the analyses. The Cognitive Concentration scale (12 items; α=.97) assessed concentration, attention, and work completion. The Authority Acceptance scale (10 items; α=.93) assessed oppositional and conduct problem behaviors (e.g., takes property, breaks rules, teases, is disobedient). The SHP includes nine items describing prosocial behaviors and emotion regulation. Items were rated on a 6-point scale and were summed to create a total score for social competence (α=.87). Finally, a teacher-rated peer-liking measure asked teachers to rate how liked each child was by his or her peers. All scores were standardized across the sample, and then a mean score was computed for each class. Missing data at either pre- or posttest reduced the sample by
Annual progress in child psychiatry and child development 2000–2001
562
5%, to 357 classrooms. The teacher-rated measure of peer liking was significantly related to peer sociometric ratings of aggression (r=.30, p<.001) and prosocial behavior (r=.34, p<.001). Peer Nominations. Peer reports of aggression, hyperactive-disruptive behavior, prosocial behavior, and likability (“most liked”) were assessed using individual sociometric interviews, which were conducted with each child whose parent had provided consent. The interviewer provided the child with a class roster and read each of the names on the roster aloud in order to ensure that the child was familiar with his or her classmates. Children were then read a series of behavioral descriptors and asked to nominate as many children as they wanted to fit each descriptor. The descriptor for aggression was “Some kids start fights, say mean things and hit other kids. Who are the kids who start fights and say mean things?” The descriptor for hyperactive-disruptive behavior was “Some kids get out of their seat a lot, do strange things, and make a lot of noise. They bother people who are trying to work. Who are the kids that get out of their seats and bother people?” The descriptor for prosocial behavior was “Some kids are really good to have in your class because they cooperate, help others, and share. They let other kids have a turn. Who are the kids who cooperate help, and share?” The descriptor for “most liked” was “Who are the kids who you like most?” The sociometric scores for each classroom were first corrected for the number of raters. Then the classroom mean scores for children were standardized across the entire sample within each cohort. Eight classrooms (approximately 2%) did not complete sociometric assessments because fewer than 70% of parents consented to the assessment. Observer Ratings. To assess the intervention effects on the high-risk children selected for the additional interventions, observers watched these children and their matched controls twice on different days for 30 minutes in their classroom (Wehby, Dodge, Valente, & CPPRG, 1993). Using a computer-assisted rating program (ASKER; Tapp & Fiel, 1991), the observers then rated ten 5-point items summarizing the classroom atmosphere during each 30-minute observation. These ratings focused on the atmosphere of the entire classroom and ranged from 1 (low) to 5 (high), with behavioral descriptions at the ends and at the midpoint. The classroom atmosphere ratings were derived from the Classroom Rating Form (Solomon, Watson, Delucchi, Schaps, & Battistich, 1988) and assessed the classroom’s: (a) level of disruption during academic time; (b) ability to handle classroom transitions; (c) ability to follow rules; (d) level of cooperation; (e) use of problem solving during conflict or need; (f) ability to express feelings appropriately; (g) level of interest and enthusiasm; (h) ability to stay focused and on task; (i) responsiveness to individual student’s needs and feelings; and (j) level of criticism versus supportiveness. Although the number of ratings per class was dependent on the number of “high-risk” children in a particular classroom, there was no difference in the number of times intervention versus control classrooms were rated; the mean number of observations was 4.4 per class. These ratings were completed in 311 classrooms; there were no high-risk intervention or control children in 77 classrooms (19%), and thus observations were not conducted in these classrooms. Six observers were trained at each site each year for approximately 6 weeks using videotapes and in situ practice sessions. To assess reliability, we randomly chose approximately 12% of the sessions to be coded by a
Initial impact of the fast track prevention trial for conduct problems
563
second observer. The kappa coefficients ranged from .62 to .81 for the 10 ratings. A mean score was derived across the number of rating occasions in each classroom. The 10 ratings had high internal consistency (a=.92); thus, a mean score for each rating and a total classroom atmosphere score were computed for each classroom. The total classroom atmosphere score (a higher score indicates poorer atmosphere) was significantly related to higher teacher ratings of problem behavior (using the Authority Acceptance scale; r=.30, p<.001), poorer Cognitive Concentration (r=.26, p<.01), low social competence (r= .28, p<.001), and higher peer-rated sociometric aggression (r=.21, p<.01).
RESULTS Statistical Models for Outcome Analyses For the teacher ratings and sociometric data, we used hierarchical linear modeling (HLM) models, with classroom as the second level in a mixed model design. Although random assignment occurred at the level of sets of schools (because the universal intervention was delivered at the classroom level), classroom was included in all analyses as a random factor using HLM (Bryk & Raudenbush, 1992). The rationale here is that classrooms, especially teachers and specific peers, may exert a sizable effect on individual scores; in fact, classrooms within schools might vary as much as classrooms across schools (Kellam, Ling, Merisca, Brown, & Ialongo, 1998). This effect might occur through influence by the teacher, classroom climate and rules, or peer group. In this case, the variation in a dependent variable across children after the effect of intervention is statistically removed, is not random, and thus violates the assumption of independence. If classrooms do indeed exert a strong effect on scores (i.e., the intraclass correlation is high), the individual-level analysis will overestimate the statistical significance of the intervention effect, whereas the classroom-level analysis will take the clustering into account to yield a lower level of statistical confidence (significance). Furthermore, the use of HLM to compute effect sizes with classroom as the unit will lead to modeling of true score relationships (the constant Level 1 measurement error is estimated and reported as sigma squared) in contrast with observed score relationships for the analysis at the individual child level only. Measurement error acts to attenuate regression coefficients, making effects appear smaller than they really are. With HLM, the coefficients are disattenuated because measurement error has been modeled as a separate parameter. The result is that, with HLM, estimates of the intervention effect will tend to be more accurate than those with individual-level analyses only. Because the high-risk intervention children also received numerous other interventions that may have affected their scores, we conducted all analyses twice, both with and without the high-risk intervention and control children included in the classroom means. Although no substantive differences were found between these two types of analyses, we present the more conservative data that excluded the high-risk intervention and control children (the highest 10% of each classroom on behavioral risk). In some cases, teachers contributed multiple cases across cohorts; in these analyses, each cohort was treated as a new case.
Annual progress in child psychiatry and child development 2000–2001
564
HLM Analyses: Sociometrics. Analyses were conducted using the HLM 4.02 software package. Full maximum likelihood estimation was used for all models. For the HLM analysis of the sociometric variable, an unconditional model was fit as the first step to examine the effects of gender (coded as −1 and 1). The output generated from this model was evaluated to determine the appropriate second step. On the basis of the recommendations of Byrk and Raudenbush (1992), Level 1 predictors were maintained if: (a) there was evidence of a significant (p<.05) fixed effect; or (b) there was evidence of significant (p<.05) Level 2 variation. Analyses included an N of 369 classrooms. The full model follows:
For aggression, at Level 1, both the fixed effect, t(368)=12, p<.001, and variance estimates, c2(1, N=369)=515, p<.001, were significant for gender, with boys showing higher scores for aggression. Thus, gender was treated as a fixed effect, as shown in the full model. The effect of intervention was significant, t(368)=2.14, p=.03; the coefficient for the intervention effect was −.05, with a standard error of .03. Intervention classrooms had lower aggression scores than did the control classrooms. Cohen’s d was −.22, as computed from the T ratio and degrees of freedom . For hyperactive-disruptive behavior, at Level 1, both the fixed effect, t(368) =−5.01, p<.001, and variance estimates, χ2(1, N=369)=522, p<.001, were significant for gender, with boys showing higher scores for hyperactive-disruptive behavior. Thus, gender was treated as a fixed effect, as shown in the full model. The effect of intervention was significant, t(368)=2.23, p=.02; the coefficient for the intervention effect was −.05, with a standard error of .02. Cohen’s d was −.22. For prosocial behavior and “most liked” ratings, the effect of intervention was not significant. There were no Site×Intervention or Cohort×Intervention interaction effects on any sociometric outcome. The intraclass correlation coefficient for sociometric variables ranged from .02 to .07. For descriptive purposes, the aggregated classroom means for sociometric variables for each group are shown in Table 30.2. Teacher Ratings of Behavior. The model for both the TOCA-R and SHP analyses was similar to the two-step model shown for sociometric analyses. However, the pretest covariate (fall of the school year) was entered as an additional Level 1 variable. There were no significant intervention or Intervention ×Other Factor effects on any teacher ratings. The intraclass correlation coefficients ranged from .08 to .21 for the teacher ratings.5 For descriptive purposes, the aggregated classroom means for teacher ratings for
Initial impact of the fast track prevention trial for conduct problems
565
each group are shown in Table 30.3.
TABLE 30.2 Peer Nomination Means and Standard Deviations for Intervention and Control Conditions (Classroom Aggregate Scores)
Variablea Aggression Hyperactive-disruptive Prosocial behavior Most liked aLower scores indicate better functioning.
Intervention M SD −0.08 0.22 −0.09 0.22 0.03 0.33 0.04 0.38
Control M SD −0.01 0.26 −0.04 0.24 0.02 0.34 0.05 0.43
TABLE 30.3 Pre- and Posttest Teacher TOCA-R Means and Standard Deviations for Intervention and Control Conditions (Classroom Aggregate Scores)
Intervention Control Pretest Posttest Pretest Posttest Variable M SD M SD M SD M SD a 0.81 0.42 0.95 0.44 0.78 0.49 0.98 0.55 Authority acceptance a 1.64 0.52 1.60 0.49 1.64 0.61 1.62 0.59 Cognitive concentration a 16.19 5.33 15.92 4.95 15.58 6.13 15.88 6.08 Low social competence b 3.71 0.69 3.77 0.59 3.91 0.69 3.87 0.67 Peer liking Note. TOCA-R=Teacher Observation of Classroom Adaptation-Revised. aHigher scores indicate greater problem behavior. bHigher scores indicate greater liking. Observer Ratings. Given that the observer ratings already occurred at the classroom level, they were analyzed using the GLM model. We conducted a 4 (Site)×2 (Intervention vs. Control) ANCOVA with cohort serving as a covariate and classroom assigned as a random effect. We then entered the Intervention× Site interaction, the Intervention×Cohort interaction, and the three way interaction of Intervention×Site×Cohort into the model. Other possible interaction effects were not entered into the model because they did not involve effects of the intervention. Analyses of the summary score of classroom atmosphere indicated a significant effect for the intervention, F(1, 279)=5.63, p<.01. Intervention classrooms were rated as having a more positive classroom atmo5
When a general linear modeling (GLM) analysis of covariance (ANCOVA) was used on mean classroom TOCA-R scores (with classroom assigned as a random effect), teachers of intervention classrooms rated their students as having a lower mean conduct problem score
Annual progress in child psychiatry and child development 2000–2001
566
(using the Authority Acceptance scale), F(1, 333)=3.68, p<.05, and a higher mean peer-liking score, F(1, 333)=3.95, p<.05, than did teachers in control classrooms. The significant effect on conduct problems indicated that during the school year, intervention classrooms showed less deterioration in behavior than did control classrooms, whereas in the area of peer liking, the intervention classrooms showed improvement relative to the control classrooms. There were no main or interaction effects on teacher ratings of cognitive competence or social competence.
sphere; the intervention classroom mean was 2.68 (SD=0.64), and the control classroom mean was 2.88 (SD=0.65). Exploratory analyses of each of the 10 separate rating scales indicated that the means favored the intervention classrooms in all cases and that 4 of the 10 subratings were statistically significant: students’ ability to follow rules, F(1, 279) =4.10, p<.05; students’ ability to express feelings appropriately, F(1, 279)=4.34, p<.05; the classroom level of interest and enthusiasm, F(1, 279)=8.21, p<.01; and the classroom’s ability to stay focused and on task, F(1, 279)=9.52, p<.005. Examination of the Role of Teacher Experience. Because three cohorts (years of classrooms) of children were used in the project, some teachers taught more than 1 year, and thus it was possible to examine the factor of teacher experience. One hundred thirteen teachers taught PATHS or control classrooms for 1 year, 47 teachers taught PATHS or control classrooms for 2 years, and 54 teachers taught PATHS or control classrooms for all 3 years. Levels of experience were not different for teachers in intervention versus control schools. All GLM and HLM analyses were redone, including years of teacher experience (coded as 1, 2, or 3) as a main effect. In GLM analyses, number of cohorts taught was used as both a main effect and an interaction effect with site and condition. In HLM analyses, number of cohorts taught was added as a Level 2 characteristic. There was only one main effect: For both the intervention and control groups, teachers who taught more cohorts had higher classroom atmosphere ratings (total score). There were no other significant main or interaction effects in these analyses, and the effects of intervention did not change. Model for Dosage and Implementation Analyses For analyses within the intervention classrooms, GLM analyses examined the relationship between dosage (highest lesson reached), quality of implementation ratings, and outcome. Outcome measures included only those found to have significant intervention versus control differences (in these analyses, we included teacher ratings found to be significant in the GLM analyses; see Footnote 5). First, analyses examined the effect of dosage with variables entered in the following order: Site and cohort were entered first as covariates, followed by the highest number of lessons taught, Highest Number of Lessons Taught×Site, Highest Number of Lessons Taught×Cohort, and the three-way interaction. Second, to examine quality of implementation, we conducted analyses separately for each of the four implementation ratings after controlling for the number of lessons taught. Thus, analyses of quality of implementation used variables entered in the following order: Site and cohort were entered first as covariates, followed by the highest number of lessons taught, implementation rating. Implementation Rating×Site, Implementation Rating×Cohort, and the three-way interaction. Interaction
Initial impact of the fast track prevention trial for conduct problems
567
effects are reported only where they alter the interpretation of the main effect of dosage or quality of implementation. Dosage. There was a trend relationship of dosage and sociometric ratings of aggression, F(1, 169)=2.61, p<.10. Higher dosage was related to somewhat lower ratings of aggression. There was also a main effect of dosage on observer ratings of classroom atmosphere, F(1, 139)=3.82, p<.05; higher dosage predicted more positive classroom atmosphere ratings. There were no main effects of dosage on the teacher ratings. Quality of Implementation. The teacher’s rated skill in teaching PATHS concepts, managing the classroom, and modeling and generalizing PATHS concepts throughout the classroom day were all significantly related to teacher ratings of Authority Acceptance: F (1, 167)=9.90, p<.001; F(1, 167)=16.54 p<.001; and F(1, 167)=9.22, p<.001, respectively. These same three measures were also significantly related to observer ratings of classroom atmosphere: F(1, 137)=4.95, p<.01; F(1, 137)=8.87, p<.01; and F(1, 137)= 4.94, p<.01, respectively. There were no main effects of these three implementation ratings on sociometric outcomes. The openness to consultation rating was related only to the classroom sociometric score of hyperactivedisruptive behavior, F (1, 168)=5.48, p<.05. In all cases, ratings of teacher skill in program implementation and classroom management predicted positive program outcomes.
DISCUSSION The results of this universal intervention model at the end of Grade 1 provide evidence of its effectiveness in the domains of both aggression and peer relations. There were significant effects of the intervention from the viewpoints of both peers and unbiased observers. Effects on aggression and classroom behavior were found from child and peer reports and the observers’ ratings (of rule following, better classroom atmosphere, and more on-task behavior). These findings reflect robust effects of the universal-level prevention activities on classroom behavior with reductions in aggression and increases in self-control and on-task behavior. This is the first reported study of a universal social competence intervention implemented at the elementary grades that used the classroom, rather than the student, as the unit of analysis. Thus, analysis occurs at the level of implementation. One previous report (Grossman et al., 1997) used the school as the unit of analysis; however, it is at the classroom level and not the school level that intervention is delivered and that intervention delivery and dosage can be reliably assessed. Given the concordant classroom-level findings for peer and observer ratings, it was suprising that teacher ratings of aggression were not found significant in the HLM analyses. To further understand this puzzling finding, we computed GLM analyses using the classroom aggregate mean score for both teacher ratings and peer sociometric data. For the measures of peer sociometrics, there was convergence between the findings of the two statistical models. This can be attributed to the relatively low intraclass correlation coefficient (Murray & Short, 1997), which demonstrated little dependency between scores of children within an average classroom. Thus, it did not appear that classroomlevel effects were exerting a significant influence on individual scores. In contrast, for
Annual progress in child psychiatry and child development 2000–2001
568
teacher ratings, there were notable differences between GLM and HLM analyses, with GLM analyses indicating a significant intervention effect on aggression/disruption (using the Authority Acceptance scale) and peer liking, but HLM models indicating no significant intervention effects. In the case of teacher ratings, the intraclass correlations were quite high (mean correlation across measures=.15), indicating high dependence between scores within classrooms. An important difference between these two measures is that peer sociometrics are rated separately by each child (in a manner similar to other measures used in many HLM models, such as academic achievement), whereas the teacher ratings are all completed by one teacher for the entire class. It is likely that this difference in having the same reporter for all class members explains the high intraclass correlation in the teacher ratings and may also account for differences between the HLM and GLM results. That is, the HLM analyses are designed to correct for the effects that children in the classroom have on each other’s behavior, removing this source of variation from the estimation of the intervention effects. In the case of teacher ratings, however, HLM corrects not only for the potential dependence between children’s actual behavior but also for the fact that teachers are likely to see their classrooms in a certain way and this perception affects the scores they provide for each child. Given the inflation of the intraclass correlation caused by using a single rater, we believe that the HLM analyses are likely to be a conservative estimate of intervention effectiveness. As two independent sources (peer ratings and observers) showed convergence in the documentation of classroom-level findings, we feel that the findings support the interpretation of a robust impact of the universal prevention activities on classroom processes. In addition, the fact that there were no significant Site×Condition interaction effects indicates no major differences in effects of intervention as a function of rural versus urban school location, percentage of children below the poverty level, or ethnic composition of the classrooms. It is unfortunate that because of the large size of the sample, we were not able to assess the acquisition of specific skills or test models in which changes in such skills might mediate outcomes. Pre-post differences in teacher ratings of classroom aggression and disrup-tion indicated an increase in both groups; that is, all teachers saw their classroom as more disruptive in the spring than they did in the previous fall. This is likely due to a combination of greater familiarity with students (and thus observation of more incidences of misbehavior) and the fact that children often show more disruption later in the classroom year than in the first few weeks (at which time they are often “on their best behavior”; Dodge, Coie, & Brakke, 1982). Although intervention effects were found across raters, these effects were modest in size. This may in part be due to the fact that this was a carefully randomized trial and within schools all teachers, independent of personal interest or ability, were assessed. Thus, teachers who were relatively ineffective, showed little enthusiasm, or completed only a portion of the curriculum were assessed as if they had completed the intervention; no intervention teachers were dropped for poor quality implementation, high resistance, or providing a low dosage. In this sense, this “intent-to-intervene” trial design (Brown, 1993) may provide the highest level of external validity regarding how this universal intervention might affect a typical, entire school community. The findings on implementation and dosage lend credence to the outcome effects. They
Initial impact of the fast track prevention trial for conduct problems
569
indicate that the intervention staffs’ ratings of how well teachers understood concepts, generalized skills outside the curriculum time, and managed their classroom were significantly related to decreases in classroom aggression (based on teachers’ mean ratings) as well as observers’ ratings of the classroom atmosphere. Further, these implementation effects continued after covarying dosage. As dosage itself had only a mild impact on outcome, it may be less crucial how many lessons are taught (given that a majority are implemented); instead, it may be the teacher’s willingness and ability to accept and use such a classroom model that most affects classroom outcomes (Elias et al., 1998). Because there have been few studies on social competence promotion at any age that have examined how dosage and quality of implementation affect outcomes (Battistich, Schaps, Watson, & Solomon, 1996; Botvin, Baker, Dusenbury, Tortu, & Botvin, 1990; Pentz et al., 1990), these findings indicate that both quantitative and qualitative indexes of curriculum implementation should be assessed in future projects. Two cautions should be raised, however, in interpreting these findings. First, because of the number of classrooms and the unique relationship of the ECs and teachers, we did not collect interrater reliability data on the implementation ratings. Thus, it is possible that these ratings are in part measuring the perception of the quality of relationship between the EC and the teacher. Second, we cannot rule out the hypotheses that teachers who are better at implementing the universal model are just better teachers in general and that the general quality of the teacher may account for these effects. However, the fact that these ratings, as hypothesized, predicted sociometric scores as well as independent classroom observations supports their validity. It should be recognized that the Fast Track universal intervention included intensive intervention with high-risk children as an integral part of the overall universal intervention. Although analyses with and without these high-risk children showed similar patterns, it is quite possible that effects of the intervention on the non-high-risk children depend on a simultaneous intervention with the high-risk children (CPPRG, 1999a). Project staff commitment to work with high-risk children may reduce teacher stress and increase teacher interest in implementing a universal intervention. Likewise, improvements in the highrisk children that are due to the selective intervention may improve classroom peer relations among other children. The present study was not designed to evaluate the effects of a universal intervention that excludes simultaneous intensive intervention with a selected group of high-risk children; thus, it does not assess the use of only the universal intervention alone. Instead, this study provides clear support that an integrated approach that combines universal and selective intervention can have meaningful effects at the universal level of analysis. These results indicate the effectiveness that a universal intervention, when implemented with fidelity and high dosage, can have in altering the quality of the classroom climate during the first year of school. It is the largest study of its kind indicating the efficacy of school-based, universal interventions during the elementary school years for both the promotion of competence (Elias, 1995) and the prevention of maladjustment (Caplan et al., 1992; Dolan et al., 1993; Grossman et al., 1997; O’Donnell, Hawkins, Catalano, Abbott, & Day, 1995). Although 1 year of preventive intervention is of value, the development and evaluation of more comprehensive models that are sustained across multiple years and grade levels are necessary to document the true
Annual progress in child psychiatry and child development 2000–2001
570
potential of such models (Weissberg & Greenberg, 1998). In the present project, the universal intervention is provided through Grade 5; in the future, we will examine the effects of multiple years of such exposure.
REFERENCES Battistich, V., Schaps, E., Watson, M., & Solomon, D. (1996). Prevention effects of the Child Development Project: Early findings from an ongoing multisite demonstration trial. Journal of Adolescent Research, 11, 12–35. Battistich, V., Solomon, D., Watson, M., Solomon, J., & Schaps, E. (1989). Effects of an elementary school program to enhance prosocial behavior on children’s cognitive social problem-solving skills and strategies. Journal of Applied Developmental Psychology, 10, 147–169. Beelmann, A., Pfingsten, U., & Losel, F. (1994). Effects of training social competence in children: A meta-analysis of recent evaluation studies. Journal of Clinical Child Psychology, 23, 260–271. Belsky, J., & MacKinnon, C. (1994). Transition to school: Developmental trajectories and school experiences. Early Education & Development, 5, 106–119. Bierman, K., Greenberg, M.T., & Conduct Problems Prevention Research Group. (1996). Social skills training in the Fast Track program. In R.D.Peters & R.J. McMahon (Eds.), Preventing childhood disorders, substance abuse, and delinquency (pp. 65–89). Thousand Oaks, CA: Sage. Botvin, G.J., Baker, E., Dusenbury, L., Tortu, S., & Botvin, E.M. (1990). Preventing adolescent drug abuse through a multimodal cognitive-behavioral approach: Results of a 3-year study. Journal of Consulting and Clinical Psychology, 58, 437–446. Brown, C.H. (1993). Statistical methods for preventive trials in mental health. Statistics in Medicine, 12, 289–300. Bryk, A.S., & Raudenbush, S.W. (1992). Hierarchical linear models. Applications and data analysis methods. Newbury Park, CA: Sage. Caplan, M., Weissberg, R.P., Grober, J.H., Sivo, P.J., Grady, K., & Jacoby, C. (1992). Social competence promotion with inner-city and suburban young adolescents: Effects on social adjustment and alcohol use. Journal of Consulting and Clinical Psychology, 60, 56–63. Cheney, D., Greenberg, M., & Kusche, C. (1991). Teacher effectiveness with PATHS: A social-cognitive curriculum for elementary students. In J.Tindal & J.Marr (Eds.), The Oregon conference monograph—1991 (pp. 81–85). Eugene, OR: University of Oregon, College of Education. Conduct Problems Prevention Research Group. (1992). A developmental and clinical model for the prevention of conduct disorders: The FAST Track program. Development and Psychopathology, 4, 509–527. Conduct Problems Prevention Research Group. (1999a). Initial impact of the Fast Track prevention trial for conduct problems: I. The high-risk sample. Journal of Consulting and Clinical Psychology, 67, 631–647. Conduct Problems Prevention Research Group. (1999b). Technical report for the Social
Initial impact of the fast track prevention trial for conduct problems
571
Health Profile. Unpublished manuscript. Denham, S.A., & Almeida, M.C. (1987). Children’s social problem solving skills, behavioral adjustment, and interventions: A meta-analysis evaluating theory and practice. Journal of Applied Developmental Psychology, 8, 391–409. Dodge, K.A., Coie, J.D., & Brakke, N.P. (1982). Behavior patterns of socially rejected and neglected preadolescents: The roles of social approach and aggression. Journal of Abnormal Child Psychology, 10, 389–409. Dolan, L.J., Kellam, S.G., Brown, C.H., Werthamer-Larsson, L., Rebok, G.W., Mayer, L.S., Laudolf, J.Ford, C., & Wheeler, L. (1993). The short-term impact of two classroom-based preventive interventions on aggressive and shy behavior and poor achievement. Journal of Applied Developmental Psychology, 14, 317345. Duriak, J.A., (1995). School-based prevention programs for children and adolescents. Thousand Oaks, CA: Sage. Elias, M.J., (1995). Primary prevention as health and social competence promotion. Journal of Primary Prevention, 16, 5–14. Elias, M.J., Gara, M.A., Schuyler, T.F., Branden-Muller, L.R., & Sayette, M.A. (1991). The promotion of social competence: Longitudinal study of a preventive school-based program. American Journal of Orthopsychiatry, 67, 409–417. Elias, M.J., Zins, J.E., Weissberg, K.S., Greenberg, M.T., Haynes, N.M., Kessler, R., Schwab-Stone, M.E., & Shriver, T.P. (1998). Promoting social and emo-tional learning: Guidelines for educators. Alexandria, VA: Association for Supervision and Curriculum Development. Greenberg, M.T., & Kusche, C.A. (1993). Promoting social and emotional development in deaf children: The PATHS project. Seattle, WA: University of Washington Press. Greenberg, M.T., & Kusche, C.A. (1997, April). Improving children’s emotion regulation and social competence: The effects of the PATHS curriculum. Paper presented at the meeting of the Society for Research in Child Development, Washington, DC. Greenberg, M.T., & Kusche, C.A. (1998). Preventive intervention for school-aged deaf children: The PATHS curriculum. Journal of Deaf Studies and Deaf Education, 3, 49– 63. Greenberg, M.T., Kusche, C.A., Cook, E.T., & Quamma, J.P. (1995). Promoting emotional competence in school-aged deaf children: The effects of the PATHS curriculum. Development and Psychopathology, 7, 117–136. Greenberg, M.T., Kusche, C.A., & Speltz, M. (1991). Emotional regulation, selfcontrol and psychopathology: The role of relationships in early childhood. In D. Cicchetti & S.Toth (Eds.), Internalizing and externalizing expressions of dysfunction: Rochester Symposium on Developmental Psychopathology (Vol. 2, pp. 21–55). New York: Cambridge University Press. Grossman, D.C., Neckerman, H.J., Koepsell, T.D., Liu, P.Y., Asher, K.N., Beland, K., Frey, K., & Rivara, F.P. (1997). The effectiveness of a violence prevention curriculum among children in elementary school. JAMA: Journal of the American Medical Association, 277, 1605–1611. Hawkins, J.D., & Weis, J.G., (1985). The Social Development Model: An integrated approach to delinquency prevention. Journal of Primary Prevention, 6, 73–97.
Annual progress in child psychiatry and child development 2000–2001
572
Institute of Medicine. (1994). Reducing risks for mental disorders: Frontiers for preventive intervention research. Washington, DC: National Academy Press. Kazdin, A.E. (1990, June). Prevention of conduct disorder. Paper presented at the National Conference on Prevention Research. National Institute of Mental Health, Bethesda, MD. Kazdin, A.E. (1993). Treatment of conduct disorder: Progress and directions in psychotherapy research. Development and Psychopathology, 5, 277–310. Kellam, S.G., Ling, X., Merisca, R., Brown, C.H., & Ialongo, N. (1998). The effect of level of aggression in the first grade classroom on the course and malleability of aggressive behavior into middle school. Development and Psychopathology, 10, 165– 186. Kendler, T.S. (1963). Development of mediating responses in children. Monographs of the Society for Research in Child Development, 28(2), 35–51. Kusche, C.A., & Greenberg, M.T. (1994). The PATHS curriculum. Seattle, WA: Developmental Research and Programs. Loeber, R. (1990). Development and risk factors of juvenile antisocial behavior and delinquency. Clinical Psychology Review, 10, 1–41. McMahon, R.J., Slough, N.M., & Conduct Problems Prevention Research Group. (1996). Family-based intervention in the Fast Track program. In R.D.Peters & R.J.McMahon (Eds.), Preventing childhood disorders, substance abuse, and delinquency (pp. 90– 110). Thousand Oaks, CA: Sage. Moffitt, T.E. (1993). Adolescence-limited and life-course persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701. Murray, D.M., & Short, B.J. (1997). Intraclass correlation among measures related to tobacco use by adolescents: Estimates, correlates, and applications to intervention studies. Addictive Behaviors, 22, 1–12. Murray, D.M., & Wolfinger, R.D. (1994). Analysis issues in the evaluation of community trials; Progress toward solutions in SAS/STAT MIXED [Special issue]. Journal of Community Psychology, 22, 140–154. O’Donnell, J., Hawkins, J.D., Catalano, R.F., Abbott, R.D., & Day, L.E. (1995). Preventing school failure, drug use, and delinquency among low-income children: Long-term intervention in elementary schools. American Journal of Orthopsychiatry, 65, 87–100. Pennington, B.F., & Welsh, M. (1995). Neuropsychology and developmental psychopathology. In D.Cicchetti & D.J.Cohen (Eds.). Developmental psychopathology: Vol. I. Theory and methods (pp. 254–290). New York: Wiley. Pentz, M.A., Trebow, E.A., Hansen, W.B., MacKinnon, D.P., Dwyer, J.H., Anderson Johnson, C., Flay, B.R., Daniels, S., & Cormack, C. (1990). Effects of program implementation on adolescent drug use behavior: The Midwestern Prevention Project. Evaluation Review, 14, 264–289. Pianta, R.C., Steinberg, M.S., & Rollins, K.B. (1995). The first two years of school: Teacher-child relationships and deflections in children’s classroom adjustment. Development and Psychopathology, 7, 295–312. Schneider, B.H. (1992). Didactic methods for enhancing children’s peer relations: A quantitative review. Clinical Psychology Review, 12, 363–382.
Initial impact of the fast track prevention trial for conduct problems
573
Solomon, D., Watson, M.S., Delucchi, K.L., Schaps, E., & Battistich, V. (1988). Enhancing children’s prosocial behavior in the classroom. American Educational Research Journal, 25, 527–554. Tapp, J.T., & Fiel, D. (1991). ASKER: A computerized program for administering Likert-scale questions [Unpublished computer program]. Nashville, TN: Vanderbilt University. Waters, E., & Sroufe, L.A. (1983). Social competence as a developmental construct. Developmental Review, 3, 79–97. Wehby, J.H., Dodge, K.A., Valente, E., & Conduct Problems Prevention Research Group. (1993). School behavior of first grade children identified as at-risk for development of conduct problems. Behavioral Disorders, 19, 67–78. Weissberg, R.P., Caplan, M.Z., & Bennetto, L. (1988). The Yale-New Haven Middle School Social Problem Solving (SPS) Program. New Haven, CT: Yale University, Department of Psychology. Weissberg, R.P., Caplan, M.Z., & Sivo, P.J. (1989). A new conceptual framework for establishing school-based social competence promotion programs. In L.A.Bond & B.E.Compas (Eds.), Primary prevention and promotion in the schools (pp. 255–296). Newbury Park, CA: Sage. Weissberg, R.P., & Elias, M.J. (1993). Enhancing young people’s social competence and health behavior: An important challenge for educators, scientists policy makers, and funders, Applied & Preventive Psychology: Current Scientific Perspectives, 3, 179– 190. Weissberg, R., & Greenberg, M.T. (1998). Community and school prevention. In I. Siegel & A.Renninger (Eds.), Handbook of child psychology: Vol. 4. Child psychology in practice (5th ed., pp. 877–954). New York: Wiley. Werthamer-Larsson, L., Kellam, S.G., & Wheeler, L. (1991). Effect of first grade classroom environment on shy behavior, aggressive behavior, and concentration problems. American Journal of Community Psychology, 19, 585–602. White, S.H. (1965). Evidence for a hierarchical arrangement of learning processes. In L.P.Lipsett & C.C.Spiker (Eds.), Advances in child development and behavior (Vol. 2, pp. 41–89). New York: Academic Press.
Permissions PART I. DEVELOPMENTAL ISSUES 1.
2.
3.
4.
5.
6.
7.
8.
9.
Horowitz, F.D. (2000). Child development and the PITS: Simple questions, complex answers, and developmental theory. Child Development, 71, 1–10. Copyright 2000 by the Society for Research in Child Development. Reprinted with permission. Posner, M.I., & Rothbart, M.K. (2000). Developing mechanisms of self-regulation. Development and Psychopathology, 12, 427–441. Copyright 2000. Reprinted with the permission of Cambridge University Press. Sroufe, L.A., Carlson, E.A., Levy, A.K., & Egeland, B. (1999). Implications of attachment theory for developmental psychopathology. Development and Psychopathology, 11, 1–13. Copyright 1999. Reprinted with the permission of Cambridge University Press. Waters, E., Merrick, S., Treboux, D., Crowell, J., & Albersheim, L. (2000). Attachment security in infancy and early adulthood: A twenty-year longitudinal study. Child Development, 71, 684–689. Copyright 2000 by the Society for Research in Child Development. Reprinted with permission. White, B.P., Gunnar, M.R., Larson, Donzella, B., & Barr, R.G. (2000). Behavioral and physiological responsivity, sleep, and patterns of daily cortisol production in infants with and without colic. Child Development, 71, 862–877. Copyright 2000 by the Society for Research in Child Development. Reprinted with permission. Gleason, T.R., Sebanc, A.R., & Hartup, W.W. (2000). Imaginary Companions of preschool children. Developmental Psychology, 36, 419–428. Copyright 2000 by the American Psychological Association. Reprinted with permission. PART II. PARENTING Collins, W.A., Maccoby, E.E., Steinberg, L., Hetherington, E.M., & Bornstein, M.H. (2000). Contemporary research on parenting: The case for nature and nurture. American Psychologist, 55, 218–232. Copyright 2000 by the American Psychological Association. Reprinted with permission. Golombok, S., Murray, C., Brinsden, P., & Abdalla, H. (1999). Social versus biological parenting: Family functioning and the socioemotional development of children conceived by egg or sperm donation. Journal of Child Psychology and Psychiatry, 40, 519–527. Copyright 1999 by the Association for Child Psychology and Psychiatry. Reprinted with permission Oyserman, D., Mowbray, C., Meares, P.A., & Firminger, K.B. (2000). Parenting among mothers with a serious mental illness. American Journal of Orthopsychiatry, 70, 296–315. Copyright 2000 by the American Orthopsychiatric Association. Reprinted with
Permissions
10.
11
12.
13.
14.
15.
16.
17.
18.
19.
575
permission. PART III. ADHD Doyle, A.E., Biederman, J., Seidman, L.J., Weber, W., & Faraone, S.V. (2000). Diagnostic efficiency of neuropsychological test scores for discriminating boys with and without attention deficit-hyperactivity disorder. Journal of Consulting and Clinical Psychology, 68, 477–488. Copyright 2000 by the American Psychological Association. Reprinted with permission. . Handen, B.L., Feldman, H.M., Lurier, A., & Huszar Murray, P.J. (1999). Efficacy of methylphenidate among preschool children with developmental disabilities and ADHD. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 805–812. Copyright 1999 by Lippincott, Williams and Wilkins. Reprinted with permission. Sharp, W.S., Walter, J.M., Marsh, W.L., Ritchie, G.F., Hamburger, S.D., & Castellanos, F.X. (1999). ADHD in girls: Clinical comparability of a research sample. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 40–47. Copyright 1999 by Lippincott, Williams and Wilkins. Reprinted with permission. Angold, A., Erkanli, A., Egger, H.L., & Costello, E.J. (2000). Stimulant treatment for children: A community perspective. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 975–984. Copyright 2000 by Lippincott, Williams and Wilkins. Reprinted with permission. PART IV. OTHER CLINICAL ISSUES Offer, D., Kaiz, M., Howard, K.I., & Bennett, E. (2000). The altering of reported experiences. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 735–742. Copyright 2000 by Lippincott, Williams and Wilkins. Reprinted with permission. Kadesjo, B., & Gillberg, C. (1999). Developmental coordination disorder in Swedish 7-year-old children. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 820–828. Copyright 1999 by Lippincott, Williams and Wilkins. Reprinted with permission. Albertini, R.S., & Phillips, K.A. (1999). Thirty-three cases of body dysmorphic disorder in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 453–459. Copyright 1999 by Lippincott, Williams and Wilkins. Reprinted with permission. Cohen, D., Flament, M, Dubos, P.F., & Basquin, M. (1999). Case series: Catatonic syndrome in young people. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 1040–1046. Copyright 999 by Lippincott, Williams and Wilkins. Reprinted with permission. Gilham, J.E., Carter, A.S., Volkmar, F.R., & Sparrow, S.S. (2000). Toward a developmental operational definition of autism. Journal of Autism and Developmental Disorders, 30, 269–278. Copyright 2000 by Kluwer Academic/Plenum Publishers. Reprinted with permission. Cyranowski, J., Frank, E., Young, E., & Shear, K. (2000). Adolescent onset of the
Annual progress in child psychiatry and child development 2000–2001
20.
21.
22.
23.
24.
25.
26.
27.
576
gender difference in lifetime rates of major depression: A theoretical model. Archives of General Psychiatry, 57, 21–27. Copyright 2000 by the American Medical Association. Reprinted with permission. Vaillant, G.E., & Davis, J.T. (2000). Social/emotional intelligence and midlife resilience in schoolboys with low tested intelligence. American Journal of Orthopsychiatry, 70, 215–222. Copyright 2000 by the American Medical Association. Reprinted with permission. PART V. TREATMENT ISSUES Angold, A., Costello, J., Burns, B., Erkanli, A., & Farmer, E.M. (2000). Effectiveness of nonresidential specialty mental health services for children and adolescents in the “real world.” Journal of the American Academy of Child and Adolescent Psychiatry, 39, 154–160. Copyright 2000 by Lippincott, Williams and Wilkins. Reprinted with permission. Erba, H.W. (2000). Early intervention programs for children with autism: Conceptual frameworks for implementation. American Journal of Orthopsychiatry, 70, 82–94. Copyright 2000 by the American Orthopsychiatric Association. Reprinted with permission. Saywitz, K.J., Mannarino, A.P., Berliner, L., & Cohen, J.A. Treatment for sexually abused children and adolescents. American Psychologist, 55, 1040–1049. Copyright 2000 by the American Psychological Association. Reprinted with permission. Silva, R.R., Munoz, D.M., Alpert, M., & Perlmutter, I.R. (1999). Neuroleptic malignant syndrome in children and adolescents, Journal of the American Academy of Child and Adolescent Psychiatry, 38, 187–194. Copyright 1999 by the Journal of the American Academy of Child and Adolescent Psychiatry. Reprinted with permission. Gutgesell, H., Atkins, D., Barst, R., Buck, M., Franklin, W., Humes, R., Ringel, R., Shaddy, R., & Taubert, K. (1999). AHA Scientific Statement: Cardiovascular monitoring of children and adolescents receiving psychotropic drugs. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 1047–1050. Originally printed in Circulation 1999, 99, 979–982 (American Heart Association). Copyright 1999 by the Journal of the American Academy of Child and Adolescent Psychiatry. Reprinted with permission. PART VI. SOCIETAL ISSUES: VIOLENCE AND VICTIMIZATION Hawker, D.S., & Boulton, M.J. (2000). Twenty years’ research on peer victimization and psychosocial maladjustment: a meta-analytic review of crosssectional studies. Journal of Child Psychology and Psychiatry, 41, 441–455. Copyright 2000 by the Association for Child Psychology and Psychiatry. Reprinted with permission. Ladd, G.W., & Burgess, K.B. (1999). Charting the relationship trajectories of aggressive, withdrawn, and aggressive withdrawn children during early grade school. Child Development, 70, 910–929. Copyright 1999 by the Society for Research in Child Development. Reprinted with
Permissions
577
permission. 28. Group for the Advancement of Psychiatry, Committee on Preventive Psychiatry. (1999). Violent behavior in children and youth: preventive intervention from a psychiatric perspective. Journal of the America Academy of Child and Adolescent Psychiatry, 38, 235–241. Copyright 1999 by Lippincott, Williams and Wilkins. Reprinted with permission. 29. Rhodes, J.E., Grossman, J.B., & Resch, N.L. (2000). Agents of change: pathways through which mentoring relationships influence adolescents’ academic adjustment. Child Development, 71, 1662–1671. Copyright 2000 by the Society for Research in Child Development. Reprinted with permission. 30. Conduct Problems Prevention Research Group. (1999). Initial impact of the fast track prevention trail for conduct problems: II. Classroom effects. Journal of Consulting and Clinical Psychology, 67, 648–657. Copyright 1999 by the American Psychological Association. Reprinted with permission.